* Logging started Sat, 01 May 2004 15:46:42 +0000
@1083426406 -dircproxy- You connected
@1083426407 -niven.freenode.net- You joined the channel
@1083426446 <housel!housel@cantor.gwydiondylan.org> I just rolled over the logfile
@1083426588 <housel!housel@cantor.gwydiondylan.org> the new one now has timestamps
@1083429853 <^self!~fn@66.209.66.105> what generates the logfile?  cia?
@1083429888 <housel!housel@cantor.gwydiondylan.org> dircproxy, which is logged in on my behalf
@1083429894 <^self!~fn@66.209.66.105> ah, ok.
@1083432027 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083434408 -niven.freenode.net- buggs^z (~noidentd@pD9E97D0D.dip.t-dialin.net) joined the channel
@1083440504 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083440504 -niven.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1083441581 <^self!~fn@66.209.66.105> http://www.livejournal.com/users/anonymous_bozo/16858.html
@1083441589 <^self!~fn@66.209.66.105> the secret is out!
@1083441682 <housel!housel@cantor.gwydiondylan.org> you don't say.
@1083442631 <^self!~fn@66.209.66.105> i'm compiling cvs/examples/ICFP2002's networking lib.  it mostly works, except one function (setnetgrent) returns void on freebsd and int on linux.  gcc bombs because of that ("void value not ignored as it ought to be").
@1083442631 <^self!~fn@66.209.66.105> how can i fix the code so that it compiles on both freebsd and linux?
@1083442714 <housel!housel@cantor.gwydiondylan.org> I usually handle that sort of thing by using my own .h file and a .c file to insulate the Dylan interface from portability issues
@1083442810 <^self!~fn@66.209.66.105> ok.
@1083442898 <^self!~fn@66.209.66.105> next problem: can i specify an a lib file to link against?  apps that use the network.lid library also need to link against -lc_r
@1083442964 <housel!housel@cantor.gwydiondylan.org> there's a lid keyword for linker options...
@1083442989 <^self!~fn@66.209.66.105> a ha.
@1083443002 <housel!housel@cantor.gwydiondylan.org> linker-options: -lc_r
@1083443279 <^self!~fn@66.209.66.105> ah.  worse than i thought.  -lc_r doesn't have what i need.
@1083443294 <housel!housel@cantor.gwydiondylan.org> what do you need?
@1083443318 <^self!~fn@66.209.66.105> gethostent_r, many other _r ones.
@1083443331 <^self!~fn@66.209.66.105> what i need to do is run melange on freebsd's networking headers.
@1083443354 <^self!~fn@66.209.66.105> i was hoping to avoid that.
@1083443397 <housel!housel@cantor.gwydiondylan.org> it's needed anyway, since linux and freebsd socket/etc structures aren't identical
@1083443474 <^self!~fn@66.209.66.105> is this bad?
@1083443475 <^self!~fn@66.209.66.105> Type name: select:1: Unknown identifier: select
@1083443475 <^self!~fn@66.209.66.105> while calling {Method rcurry{<function>} internal method ()} with #["select"]
@1083443523 <^self!~fn@66.209.66.105> ok, fixed it
@1083443532 <^self!~fn@66.209.66.105> i included sys/time.h and unistd.h
@1083447174 <^self!~fn@66.209.66.105> it works with mod_lisp
@1083447266 <^self!~fn@66.209.66.105> it's a separate package
@1083447355 <^self!~fn@66.209.66.105> er
@1083447455 <housel!housel@cantor.gwydiondylan.org> andreas's Dylan network library works with mod_lisp?
@1083447610 <^self!~fn@66.209.66.105> no, i meant to write that to #lisp.
@1083447614 <^self!~fn@66.209.66.105> but yes, i was trying to build the network library so i could get it to work with mod_lisp.  i got the network library to compile, but then ran into trouble getting some simple code to compile.
@1083448336 <^self!~fn@66.209.66.105> mod_lisp's protocol is pretty simple.  whenever it wants to send you something, it ends you key/value pairs, "key\nvalue\n...end".  you reply with key/value pairs, including a content-length, and then the data.
@1083448731 <brucehoult!~bruce@202.0.59.55> did you guys see my little experiemnt to see how much d2c builds rely on fast disks?
@1083448985 <CIA-2!micah@hyperreal.info> commit by alexpotanin to libraries: Job: 6994
@1083448985 <CIA-2!micah@hyperreal.info> Feeble attempts at making it work on NetBSD with its own /usr/pkg include hierarchy.
@1083450024 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083451356 -niven.freenode.net- hannes (~hannes@pD9E7B33C.dip.t-dialin.net) joined the channel
@1083451803 <^self!~fn@66.209.66.105> well, got to stage 1: getting the code to work on freebsd.  my dumb irc bot compiles and runs.
@1083451939 -niven.freenode.net- D-BOT (~silencio@203.215.182.99) joined the channel
@1083451953 <D-BOT!~silencio@203.215.182.99> yo.
@1083451982 <^self!~fn@66.209.66.105> heh, it doesn't like quitting:
@1083451990 <^self!~fn@66.209.66.105> C quit :dive dive dive!
@1083451990 <^self!~fn@66.209.66.105> S ERROR :Closing Link: D-BOT (Client Quit)
@1083451990 <^self!~fn@66.209.66.105> Bad file descriptor
@1083451990 <^self!~fn@66.209.66.105> Abort trap (core dumped)
@1083452091 <hannes!~hannes@pD9E7B33C.dip.t-dialin.net> hi
@1083452209 <^self!~fn@66.209.66.105> hi hannes.
@1083452249 <^self!~fn@66.209.66.105> ok, i'll try to get this modlisp thing working today.
@1083452254 <^self!~fn@66.209.66.105> for now, must slep.  night,.
@1083452308 <^self!~fn@66.209.66.105> sleep, even.
@1083454298 -niven.freenode.net- hannes_ (~hannes@pD9E7AA28.dip.t-dialin.net) joined the channel
@1083458434 [housel!housel@cantor.gwydiondylan.org] ACTION experiments with using FunDev with MPS on "linux"
@1083458458 <brucehoult!~bruce@202.0.59.55> aka FreeBSD?
@1083458477 <housel!housel@cantor.gwydiondylan.org> yes (in linuxulator mode)
@1083460509 -niven.freenode.net- dasds (sdaa@ACC5819A.ipt.aol.com) joined the channel
@1083460699 -niven.freenode.net- dasds (sdaa@ACC5819A.ipt.aol.com) left the channel
@1083462553 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083462863 -niven.freenode.net- hannes_ (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083469295 -niven.freenode.net- brucehoult (~bruce@202.0.59.134) joined the channel
@1083469610 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hmm... Over the past day I compiled Dylan (successfully) on NetBSD and Windows XP and SuSE Linux - all using the latest CVS. Unfortunately, every single platform required hacks to make it work (albeit small ones) and both NetBSD and SuSE Linux required me to configure it with "--disable-shared" for Dylan to compile... OpenGL didn't compile on either NetBSD or Windows XP, but I had it working on SuSE Linux last weekend... Frustrating... :-( But at l
@1083469682 <brucehoult!~bruce@202.0.59.134> cross platform is a reality these days, but only with a lot of hard work :-(
@1083469704 <brucehoult!~bruce@202.0.59.134> and they are all moving targets
@1083469882 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083470018 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Are you at work or home, by the way, Bruce?
@1083470029 <brucehoult!~bruce@202.0.59.134> home
@1083470058 <brucehoult!~bruce@202.0.59.134> it's Sunday, dude
@1083470093 <brucehoult!~bruce@202.0.59.134> and I quit.  Why would I be at work on a Sunday after I gave notice? ;-)
@1083470223 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> But you will be during weekdays, right? For the next couple of weeks?
@1083470232 <brucehoult!~bruce@202.0.59.134> yes
@1083470247 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Also, does it mean that you gliding spending will increase now, unless we have crap days like today?
@1083470293 <brucehoult!~bruce@202.0.59.134> possibly
@1083470306 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Will you get Instructor Rating before I get my B certificate? Or even A one - since I am about to go overseas for 5 months and won't be signing much off (apart from some B certificate stuff) till I come back... :-D
@1083470327 <brucehoult!~bruce@202.0.59.134> who knows? :)
@1083470386 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I am working right now anyway... Earning my $450 for marking one of the assignments for 26 people... That's roughly $40 a head (all after tax)... :-P
@1083470408 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I hate marking... Mind numbing activity at times... :-D
@1083470417 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> But stomach filling... ;-)
@1083470424 <brucehoult!~bruce@202.0.59.134> sounds as if the hourly rate is pretty good though
@1083470452 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Yes, it took me 3 hours to mark 10 or 11 fully... :-) And it accelerates as I go through them... :-)
@1083475881 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083478370 -niven.freenode.net- brucehoult (~bruce@202-0-60-22.adsl.paradise.net.nz) joined the channel
@1083481115 -niven.freenode.net- brucehoult (~bruce@202-0-60-84.adsl.paradise.net.nz) joined the channel
@1083482239 -niven.freenode.net- brucehoult_ (~bruce@202-0-61-17.adsl.paradise.net.nz) joined the channel
@1083483608 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083483753 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083484329 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083486167 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083486355 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083506416 -niven.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1083506417 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1083513955 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083517188 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083520799 -niven.freenode.net- buggs|afk (~noidentd@pD9E967C2.dip.t-dialin.net) joined the channel
@1083524018 -niven.freenode.net- gabor_ (root@cantor.gwydiondylan.org) joined the channel
@1083524023 <gabor_!root@cantor.gwydiondylan.org> re.
@1083525188 <^self!~fn@66.209.66.105> hi
@1083525546 <gabor_!root@cantor.gwydiondylan.org> rehi self
@1083526329 -niven.freenode.net- hannes (~hannes@217.231.170.40) joined the channel
@1083529268 -niven.freenode.net- gabor_ (root@cantor.gwydiondylan.org) joined the channel
@1083536406 <bhoult!~bhoult@morannon.eservglobal.co.nz> hey Gabor!
@1083536417 <bhoult!~bhoult@morannon.eservglobal.co.nz> oops
@1083536422 <bhoult!~bhoult@morannon.eservglobal.co.nz> several hours ago :-(
@1083536426 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION sleeps again
@1083538740 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083539933 <CIA-2!micah@hyperreal.info> commit by alexpotanin to libraries: Job: 6994
@1083539933 <CIA-2!micah@hyperreal.info> Made melange more verbose.
@1083540102 <CIA-2!micah@hyperreal.info> commit by alexpotanin to libraries: Job: 6994
@1083540102 <CIA-2!micah@hyperreal.info> Sorry, now only verbose on NetBSD - where it is necessary, unfortunately. Otherwise - segfaults for no apparent reason.
@1083540173 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083540180 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hello, OpenGL hackers. :-)
@1083540191 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> "/usr/include/GL/glext.h:2423: Processing define "GL_ACTIVE_STENCIL_FACE_EXT"make: *** [opengl-glut-intr.dylan] Error 139"
@1083541553 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083542752 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083545243 -niven.freenode.net- gabor_ (root@cantor.gwydiondylan.org) joined the channel
@1083545251 <gabor_!root@cantor.gwydiondylan.org> re.
@1083548412 -niven.freenode.net- doublec (~doublec@coretech.co.nz) joined the channel
@1083549621 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net) joined the channel
@1083551965 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Has anybody had recent contact with Jeremy Jones or Gail Zacharias? Do they know about FunDev and the activities of the Maintainers? I'm just curious how aware the old Harlequin and FunDev people are about what's going on in the Dylan universe.
@1083552070 <doublec!~doublec@coretech.co.nz> I've not heard from anyone from fundev in a long time.
@1083552118 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> I've had an email exchange with Andrew Shalit (who asked me about Jeremy and Gail, which is why I asked). He's interested in seeing the DRM updated and corrected, and even says he'd love to participate if he has time. He's also interested in trying d2c on Mac OS.
@1083553831 <bhoult!~bhoult@morannon.eservglobal.co.nz> Gail is a member of the Dylan community on Orkut
@1083553914 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. and so is Jeremy
@1083553995 <bhoult!~bhoult@morannon.eservglobal.co.nz> and Oliver Steele, who's name I'm sure I should recognize
@1083554148 <doublec!~doublec@coretech.co.nz> now building web stuff at laszlo.
@1083572250 -niven.freenode.net- pritlove (~pritlove@194.95.203.218) joined the channel
@1083572263 -niven.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1083572266 <andreas!andreas@cantor.gwydiondylan.org> Moin!
@1083572391 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> howdy
@1083572409 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Math involving a <general-integer> will never result in an <integer>, right?
@1083572449 <andreas!andreas@cantor.gwydiondylan.org> Not in GD right now.
@1083572454 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> To get an <integer>, do I just use as(<integer>, some-general-integer) ?
@1083572486 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> (Assuming the value fits in an <integer>.)
@1083572532 <andreas!andreas@cantor.gwydiondylan.org> SHould do it.
@1083573813 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083573855 -niven.freenode.net- brucehoult (~bruce@202-0-60-138.adsl.paradise.net.nz) joined the channel
@1083573877 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hi Bruce. :-)
@1083573890 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Let's see if i can get libopengl going on SuSE again... :-)
@1083573893 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> hi Alex
@1083574022 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> What exactly is melange and where does it come from?
@1083574029 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Do you compile it as part of dylan?
@1083574046 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> yes
@1083574070 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> it and parsergen are built during the bootstrap process
@1083574096 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> see the src/tools directory
@1083574122 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> OK.
@1083574141 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> How do I tell it to run gcc in the lowest optimisation level?
@1083574148 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> melange is, of course, a spice
@1083574272 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> What exactly is its generic purpose?
@1083574299 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> ?
@1083574314 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> making Dylan interfaces to C code
@1083574346 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Automatically from "special" Dylan code?
@1083574384 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> in many cases it can do it completely automatically, just by saying in the .intr file which C header to include
@1083574394 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> In other cases you need to fine-tune the process a bit
@1083574462 <brucehoult!~bruce@202-0-60-138.adsl.paradise.net.nz> amazingly, there is documentation: http://gwydiondylan.org/gdref/gduser.html#MELANGE
@1083574485 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> OK.
@1083574570 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Now, I stopped talking to Dave and now 100% doing this. :-)
@1083574588 [cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net] ACTION is working on Melange as we speak, btw
@1083574610 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Oh?
@1083574616 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Trying to add support for casting from unsigned to signed.
@1083574626 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hi Chris! Is it related to our problems, by any chance?
@1083574639 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I see. Cool.
@1083574640 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> What problems?
@1083574687 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Making libraries/libopengl compile on Linux/Windows/NetBSD or something... :-)
@1083574712 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Well, what problem, specifically?
@1083574721 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> It has five files:  opengl-exports.dylan and opengl-macros.dylan in dylan...
@1083574732 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> And three remaining: as .intr.
@1083574746 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> It uses melange to produce three dylan files for those integrating information in GL headers.
@1083574757 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Melane segfaults on all three platforms mentioned above... :-(
@1083574792 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Andreas may have it working for him on Debian - I can't confirm that.... :-)
@1083574864 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> I don't think it's related. I'm adding support for casting unsigned values to signed integer types. Melange signals an exception if it encounters an enum value like "(long)0xFFFFFFFF".
@1083574929 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I was just describing the problem we are having... :) I am not sure why it happens yet... :-(
@1083574952 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> I'm having my own problems, though, trying to fix that. With my changes, Melange signals an exception, but I can't find the generated C code to figure out exactly what it's trying to do.
@1083574979 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Ugh... :(
@1083575792 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Well, I found the generated C code, but it's a complete mystery to me. I don't recognize anything from the Dylan source.
@1083575908 <andreas!andreas@cantor.gwydiondylan.org> Hi Alex, Bruce!
@1083576026 -niven.freenode.net- brucehoult_ (~bruce@202-0-61-49.adsl.paradise.net.nz) joined the channel
@1083576076 <andreas!andreas@cantor.gwydiondylan.org> Re Bruce.
@1083576230 <brucehoult_!~bruce@202-0-61-49.adsl.paradise.net.nz> hi Andreas
@1083576355 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hi Andreas!
@1083576429 [alexpotanin!~alex@st-james.mcs.vuw.ac.nz] ACTION suspects that melange gets lost processing glu.h and glut.h and goes into infinite loop or something... I hope I am wrong... :-)
@1083576441 <andreas!andreas@cantor.gwydiondylan.org> alex: the segfaults you see in melange are suspicious. It shouldn't segfault. Could you attach a debugger and get a stack trace?
@1083576451 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> are you parsing the native include files?
@1083576459 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> alex@linux:~> ulimit -a
@1083576459 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> core file size        (blocks, -c) 0
@1083576460 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> data seg size         (kbytes, -d) unlimited
@1083576460 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> file size             (blocks, -f) unlimited
@1083576460 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> max locked memory     (kbytes, -l) unlimited
@1083576460 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> max memory size       (kbytes, -m) unlimited
@1083576462 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> open files                    (-n) 1024
@1083576464 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> pipe size          (512 bytes, -p) 8
@1083576466 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> stack size            (kbytes, -s) unlimited
@1083576468 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> cpu time             (seconds, -t) unlimited
@1083576470 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> max user processes            (-u) 4094
@1083576472 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> virtual memory        (kbytes, -v) unlimited
@1083576474 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I am parsing /usr/include/GL - presumably native. :-)
@1083576475 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> we were going to put our custom ones back, as the easier option for now
@1083576487 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Better fix this maybe? :)
@1083576493 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> well, that's why you're not getting core files, anyway :-)
@1083576501 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I can set core file to something then. :)
@1083576504 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> How?
@1083576512 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> ulimit -c nnnn
@1083576520 <andreas!andreas@cantor.gwydiondylan.org> alex: "unlimited" actually means "system limit".
@1083576557 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> nnnn in what?
@1083576563 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> blocks
@1083576570 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> or put "unlimited" :-)
@1083576574 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> OK.
@1083576578 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> (without quotes)
@1083576593 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> done
@1083576617 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Doing glu
@1083576622 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> gl.h
@1083576626 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> glext.h
@1083576629 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> glext.g again
@1083576641 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> coming up to 2423
@1083576643 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> 139.
@1083576646 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> crash. :)
@1083576653 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> core - 7MB
@1083576654 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> :)
@1083576678 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> ulimit -Ha
@1083576689 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> (gdb) bt
@1083576689 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> #0  0x0819bfed in ?? ()
@1083576689 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Cannot access memory at address 0xbfe00fd0
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Core was generated by `melange -v -I/usr/X11R6/include --d2c opengl-glu.intr ,opengl-glu-intr.dylan'.
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Program terminated with signal 11, Segmentation fault.
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> #0  0x0819bfed in ?? ()
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> (gdb) bt
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> #0  0x0819bfed in ?? ()
@1083576707 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Cannot access memory at address 0xbfe00fd0
@1083576721 <andreas!andreas@cantor.gwydiondylan.org> Eek!
@1083576727 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> What is -Ha?
@1083576733 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> hard limits
@1083576750 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Still - unlimited for all but open files, pipe and max uproc.
@1083576814 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> The thing is that glu one goes through gl.h, then glext.h and dies.
@1083576824 <brucehoult!~bruce@202-0-61-49.adsl.paradise.net.nz> I'm sure Andreas will sort you out ... I've just been invited to have an early night...
@1083576826 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> But opengl.intr works and goes through gl.h and glext.h and works. :-(
@1083576837 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> OK...
@1083576841 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Good night! :-)
@1083576851 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> By the way, Andreas - you are in Berlin, right?
@1083576864 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I will be around this summer... :-)
@1083577131 <andreas!andreas@cantor.gwydiondylan.org> Alex, I am! When will you be here?
@1083577265 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Only briefly. I am in Oslo: 11th of June to 20th of June.
@1083577276 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> In Amsterdam: 20th of June till 9th of July (?)
@1083577285 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> In Berlin after that (catching a train).
@1083577290 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Then going to Jatez.
@1083577302 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Leaving Europe for USA - 21st of 22nd of July.
@1083577317 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I am flexible as to when I go from Amsterdam to Berlin and from Berlin to Jatez... :-)
@1083577365 <andreas!andreas@cantor.gwydiondylan.org> I see. I need to make a plan for the summer... so many dates!
@1083577380 <andreas!andreas@cantor.gwydiondylan.org> But you'll be here when the Love Parade won't happen this year.
@1083577411 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> That's a good thing, I guess... :-)
@1083577424 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> How do I comment out lines in .lid file? //?
@1083577447 <andreas!andreas@cantor.gwydiondylan.org> Prefix the header line with X-.
@1083577473 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> What does X- do?
@1083577489 <andreas!andreas@cantor.gwydiondylan.org> Nothing, it just changes the semantics to "do nothing".
@1083577506 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> OK... Thanks... :)
@1083577592 <andreas!andreas@cantor.gwydiondylan.org> BTW: I just recompiled opengl, just to make sure that it still works for me.
@1083577662 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> From completely clean start nothing crashed and used glu.h and glut.h from your Debian unstable?
@1083577722 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Well, for fun's sake. I compiled the library by removing glu and glut... :-D
@1083577730 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> called it opengl-simple.lid... :-D
@1083577790 <andreas!andreas@cantor.gwydiondylan.org> alex: yep, including "make clean" and re-melanging everything.
@1083577882 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> what version of gcc and gl do you have?
@1083577908 <andreas!andreas@cantor.gwydiondylan.org> gcc version 3.3.3 (Debian 20040401)
@1083577961 <andreas!andreas@cantor.gwydiondylan.org> ii  xlibmesa-gl    4.3.0-7        Mesa 3D graphics library [XFree86]
@1083578045 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I have 3.3.1 (SuSE)...
@1083578056 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I have ATI drivers for libGL and stuff...
@1083578068 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> 4.3 x server... :)
@1083578081 <andreas!andreas@cantor.gwydiondylan.org> Hmmmm.... should be the very same headers.
@1083578102 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I guess so - must be g++ then?
@1083578117 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Can I test melange in a better way?
@1083578117 <andreas!andreas@cantor.gwydiondylan.org> g++?
@1083578117 <^self!~fn@66.209.66.105> sure.
@1083578134 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I mean the way g++ compiled melange or something like that... :-)
@1083578157 <andreas!andreas@cantor.gwydiondylan.org> Yes. gcc, or the garbage collector.
@1083578182 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I think I have gc 6.2 - whichever latest stable... :)
@1083578218 <andreas!andreas@cantor.gwydiondylan.org> I have 6.3alpha4, but I don't think there are that many changes.
@1083578287 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I would've though it be more stable in 6.2... :)
@1083578298 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I guess I can upgrade g++ and see if it fixes it...
@1083578310 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Otherwise - I can't figure out why melange segfaults... :-(
@1083578328 <andreas!andreas@cantor.gwydiondylan.org> With a broken stack trace, this is as hard as it gets.
@1083578431 <andreas!andreas@cantor.gwydiondylan.org> Just to make doubly sure, I'm now using the melange from the latest Debian packet, instead of the self-compiled one. Again, works as expected.
@1083578498 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I'll try on NetBSD now.. :)
@1083578652 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> I'm having trouble coming up with code for casting from unsigned to signed integer. Any suggestions for how you'd do it?
@1083578863 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> I may just not understand <general-integer> correctly, but I keep getting a result that won't fit in an <integer>.
@1083578902 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Andreas: make opengl-glu-intr.dylan works on NetBSD... :-) Segfaults on SuSE... :-(
@1083578934 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> What I'm doing right now is testing whether the value is larger than the largest signed value possible for the given number of bytes: value > (2 ^ (8 * int-size - 1) - 1)
@1083578992 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> Then, I try to convert from unsigned to signed like this: as(<integer>, value - 2 ^ (8 * int-size))
@1083579031 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> The conversion calculation seems to be incorrect. Maybe I'm too tired to be coding right now, but I can't tell what's wrong.
@1083579088 <andreas!andreas@cantor.gwydiondylan.org> Chris: Hm, looks correct to me on first glance.
@1083579104 <andreas!andreas@cantor.gwydiondylan.org> alex: What gcc version on NetBSD?
@1083579120 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> 3.3.3
@1083579132 <andreas!andreas@cantor.gwydiondylan.org> I'd blame gcc.
@1083579137 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> barretts: [~/ELVISFiles/Dylan/CVSData/libraries/libopengl-dylan] % make opengl-glu-intr.dylan
@1083579137 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> melange -I/usr/pkg/include -I/usr/X11R6/include --d2c opengl-glu.intr ,opengl-glu-intr.dylan && mv ,opengl-glu-intr.dylan opengl-glu-intr.dylan
@1083579138 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> barretts: [~/ELVISFiles/Dylan/CVSData/libraries/libopengl-dylan] % make opengl-glut-intr.dylan
@1083579138 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> melange -I/usr/pkg/include -I/usr/X11R6/include --d2c opengl-glut.intr ,opengl-glut-intr.dylan && mv ,opengl-glut-intr.dylan opengl-glut-intr.dylan
@1083579138 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Type name: exit:1: Unknown identifier: exit
@1083579140 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> while calling {Method rcurry{<function>} internal method ()} with #["exit"]
@1083579142 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> melange: $GLUT-NORMAL has multiple definitions
@1083579144 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> melange: $GLUT-OVERLAY has multiple definitions
@1083579146 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> barretts: [~/ELVISFiles/Dylan/CVSData/libraries/libopengl-dylan] % g++ --version
@1083579148 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> g++ (GCC) 3.3.2 (NetBSD nb1 20031026)
@1083579150 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Copyright (C) 2003 Free Software Foundation, Inc.
@1083579152 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> This is free software; see the source for copying conditions.  There is NO
@1083579154 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
@1083579158 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> barretts: [~/ELVISFiles/Dylan/CVSData/libraries/libopengl-dylan] %
@1083579160 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> barretts: [~/ELVISFiles/Dylan/CVSData/libraries/libopengl-dylan] %
@1083579162 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Sorry.
@1083579164 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> So - it fails for glut and it is 3.3.2 that does it - not 3.3.3 :-(
@1083579268 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Sorry - it doesn't quite fail - it produces .dylan file... :-(
@1083579290 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> By the way, on my Mandrake with 3.3.2 I got gc crashing on its own checks... :-(
@1083579303 <cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net> andreas: Okay, thanks. At least I'm not completely off base.
@1083579326 [cpage!~Chris@adsl-67-118-119-143.dsl.snfc21.pacbell.net] ACTION must sleep before wrestling with code any more tonight
@1083579392 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Good night, Chris!
@1083579851 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I think I will go now. I will try again once I upgrade my g++... :-) Good night!
@1083579867 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083594173 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083594174 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1083597861 -dircproxy- You disconnected
@1083599721 -dircproxy- You connected
@1083601567 -niven.freenode.net- hannes_ (~hannes@pD9E7AC10.dip.t-dialin.net) joined the channel
@1083601676 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1083601774 <hannes_!~hannes@pD9E7AC10.dip.t-dialin.net> re
@1083607200 -niven.freenode.net- buggs^z (~noidentd@pD9E9669B.dip.t-dialin.net) joined the channel
@1083610152 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083611070 -niven.freenode.net- mattsd (~mattsd@wks1.qa.mirrorimage.net) joined the channel
@1083611075 -niven.freenode.net- mattsd (~mattsd@wks1.qa.mirrorimage.net) left the channel
@1083613675 -dircproxy- You disconnected
@1083614853 -dircproxy- You connected
@1083619949 -niven.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083620446 -niven.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1083620447 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1083620825 -dircproxy- You disconnected
@1083621062 <bhoult!~bhoult@morannon.eservglobal.co.nz> re Rob
@1083626758 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is there syntax for <extended-integer> literals?
@1083626883 -niven.freenode.net- gabor_ (root@cantor.gwydiondylan.org) joined the channel
@1083626909 <gabor_!root@cantor.gwydiondylan.org> cpage: yes, there is a special syntax. one moment...
@1083627077 <gabor_!root@cantor.gwydiondylan.org> ooops, 0.0x0 is for extended float, /me still looking
@1083627137 <gabor_!root@cantor.gwydiondylan.org> maybe #e2 ?
@1083627299 <bhoult!~bhoult@morannon.eservglobal.co.nz> hey, Gabor!
@1083627333 -niven.freenode.net- hannes_ (~hannes@pD9E7988D.dip.t-dialin.net) joined the channel
@1083627348 <gabor_!root@cantor.gwydiondylan.org> hi bruce!
@1083627355 <gabor_!root@cantor.gwydiondylan.org> hi hannes!
@1083627553 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah. I was trying to use #e with hex digits.
@1083627653 <gabor_!root@cantor.gwydiondylan.org> anyone read Alistair Cockburn books?
@1083627706 <gabor_!root@cantor.gwydiondylan.org> for a talk at work I am currently diving in "Agile SW Dev."
@1083628746 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083628748 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1083630063 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Hmm. Apparently, <extended-integer> isn't contagious enough. I have to use an extended integer on the left-hand-side of an operator to get an extended result.
@1083630097 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Or, at least, on the left-hand-side of exponentiation ^.
@1083630120 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is that a feature or a bug?
@1083630186 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (ie., the right-hand-side is an extended integer, AFAIK.)
@1083630371 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083631927 -niven.freenode.net- buggs|afk (~noidentd@pD9E965D2.dip.t-dialin.net) joined the channel
@1083633225 -niven.freenode.net- cpage (~Chris@charger.intuit.com) joined the channel
@1083633403 -niven.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083633455 -niven.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083633460 -niven.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) left the channel
@1083635282 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083636532 <CIA-2!micah@hyperreal.info> commit by alexpotanin to src: Job: 6994
@1083636532 <CIA-2!micah@hyperreal.info> I fixed NetBSD platform specification to make it possible to compile libopengl.
@1083636855 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Dylan *and* libopengl work properly on NetBSD now. I am happy about one platform out of three I use 100% of the time. :-)
@1083637082 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> I am building g++ v3.4.0 on my SuSE Linux to check if it solves my problems there. :-)
@1083637858 -dircproxy- You connected
@1083642183 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083646763 <CIA-2!micah@hyperreal.info> commit by alexpotanin to libraries: Job: 6994
@1083646763 <CIA-2!micah@hyperreal.info> OK. With G++ v3.4.0 recompiled Dylan, libopengl now compiles on SuSE Linux.
@1083647037 <bhoult!~bhoult@morannon.eservglobal.co.nz> yay
@1083657835 -niven.freenode.net- brucehoult (~bruce@202-0-60-41.adsl.paradise.net.nz) joined the channel
@1083663293 -niven.freenode.net- pritlove (~pritlove@194.95.203.213) joined the channel
@1083669759 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083669962 -niven.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083670160 <alexpotanin!~alex@st-james.mcs.vuw.ac.nz> Hello. I was wondering if Bruce was around, as I usually do... :-) Or Andreas? :-)
@1083670715 <cgay!~cgay@ita4fw1.itasoftware.com> does anyone know how to setup the "personal registry root" that HughG spoke of?
@1083671175 <cgay!~cgay@ita4fw1.itasoftware.com> ah, found it in some old mail he sent
@1083671553 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: i know.
@1083671693 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: if you've any problems, feel free to ask me.
@1083671824 <cgay!~cgay@ita4fw1.itasoftware.com> hannes: i'm playing around with 'set personal-root /foo'
@1083671878 <cgay!~cgay@ita4fw1.itasoftware.com> did you have to do the trick Hugh mentioned of deleting all the registry entries for system projects?
@1083671944 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083672055 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: no. i set up my personal registry in registry/Sources/registry/x86-linux. and i do 'set personal-root 'registry/' in minimal-console-compiler
@1083672199 <cgay!~cgay@ita4fw1.itasoftware.com> ok, i'll give it a try
@1083672414 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> in registry/Sources/registry/x86-linux are those files starting with abstract://dylan/ and then the relative path starting from registry/Sources/
@1083672496 <cgay!~cgay@ita4fw1.itasoftware.com> did you set up entries to build koala?  can you just have the urls point to libraries/koala like this:  abstract://dylan/../../../libraries/koala?
@1083672555 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> yes, i have entries for building koala, but i have abstract://dylan/../../libraries/koala/sources/koala/koala-unix.hdp
@1083672565 <cgay!~cgay@ita4fw1.itasoftware.com> (i notice there's some brokenness if you first set the personal root to an absolute path and then try to set it to a relative path.  fundev gets confused.
@1083672567 <cgay!~cgay@ita4fw1.itasoftware.com> )
@1083672590 <cgay!~cgay@ita4fw1.itasoftware.com> ok, cool.  (yeah, the number of ..'s was just a guess :)
@1083672591 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> it gets confused if you set the personal root more than once
@1083672639 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> and the personal-root is lost after doing sth else (like open koala and build)
@1083672660 <cgay!~cgay@ita4fw1.itasoftware.com> so you have to set it every time?
@1083672724 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> every time i start fundev, but then i use another shell for editing koala source and run build several times
@1083672764 <cgay!~cgay@ita4fw1.itasoftware.com> k
@1083672783 <cgay!~cgay@ita4fw1.itasoftware.com> and i assume i need one registry entry for each koala subproject
@1083672896 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> yes, one file for each subproject
@1083673022 <cgay!~cgay@ita4fw1.itasoftware.com> thanks, you've saved me a lot of time :)
@1083673116 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> np
@1083673225 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> i just found a readme.unix where i wrote down how to compile koala on unix? i can mail it to you.
@1083673253 <cgay!~cgay@ita4fw1.itasoftware.com> that would be great.  cgay@itasoftware.com
@1083673389 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> sent
@1083673588 <cgay!~cgay@ita4fw1.itasoftware.com> should the SIGPIPE mods be checked into cvs?  or are they just a temporary hack?
@1083674059 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> they're a temporary hack.
@1083674480 <cgay!~cgay@ita4fw1.itasoftware.com> argh.  now i can't even bootstrap 
@1083674499 <cgay!~cgay@ita4fw1.itasoftware.com> i get the less-than-informative error: make: *** [Bootstrap.1/bin/minimal-console-compiler] Error 255
@1083674546 <cgay!~cgay@ita4fw1.itasoftware.com> i bet it's stashed something in an environment variable somewhere and still thinks i have a bootstrap.1 compiler to use
@1083674742 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> what did you change? only the patch and then a make?
@1083674781 <cgay!~cgay@ita4fw1.itasoftware.com> and the registry stuff...
@1083674838 <cgay!~cgay@ita4fw1.itasoftware.com> no, i guess nothing changed in the system registries
@1083675212 <cgay!~cgay@ita4fw1.itasoftware.com> i c.  when you call 'configure' you can't use a relative path
@1083675226 <cgay!~cgay@ita4fw1.itasoftware.com> i.e., ../configure fails.  $SRCDIR/configure works
@1083675519 <CIA-2!micah@hyperreal.info> commit by cgay to fundev README.txt,1.8,1.9: job: minor
@1083675519 <CIA-2!micah@hyperreal.info> Add a warning not to use relative path when calling configure.
@1083678896 <cgay!~cgay@ita4fw1.itasoftware.com> hannes: did you have this problem?:
@1083678900 <cgay!~cgay@ita4fw1.itasoftware.com> 2004-05-04 09:51:53 -0-4:00 [INFO] Ready for service on port 80
@1083678900 <cgay!~cgay@ita4fw1.itasoftware.com> socket error
@1083678900 <cgay!~cgay@ita4fw1.itasoftware.com> zsh: 19212 trace trap  bin/koala-app
@1083678900 <cgay!~cgay@ita4fw1.itasoftware.com> exit status: 133
@1083678917 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: you run koala as root?
@1083678925 <cgay!~cgay@ita4fw1.itasoftware.com> ah, no
@1083678942 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> i use a config file which starts koala on port 8000 ;)
@1083678994 <cgay!~cgay@ita4fw1.itasoftware.com> k
@1083679187 <cgay!~cgay@ita4fw1.itasoftware.com> woo hoo!  got my first response (404)
@1083679588 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> :)
@1083682405 <CIA-2!micah@hyperreal.info> commit by cgay to libraries: job: 6996
@1083682405 <CIA-2!micah@hyperreal.info> Document how to build and run on Linux.  Other minor updates.
@1083691951 -niven.freenode.net- hannes (~hannes@pD9E7988D.dip.t-dialin.net) joined the channel
@1083691953 <hannes!~hannes@pD9E7988D.dip.t-dialin.net> re
@1083694382 -niven.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1083694383 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1083696210 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1083696374 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Interesting: Melange seems to have a bug where it gets a parser error if an enum value is a character literal and there is an apostrophe in a block comment next to it, e.g.: myEnum = 'a' /* ' */
@1083696470 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If I change it to a // line comment, it's fine.
@1083696494 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/seems to have/has
@1083699983 -niven.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1083703877 -dircproxy- You disconnected
@1083703887 -dircproxy- You connected
@1083704094 <housel!housel@cantor.gwydiondylan.org> http://slashdot.org/comments.pl?sid=106332&cid=9051932
@1083705562 <bhoult!~bhoult@morannon.eservglobal.co.nz> heh .. I've got 5 mod points -- I could mod you down to zero...
@1083706225 -niven.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1083706303 <cpage_!~Chris@charger.intuit.com> Note to self: Debugging Melange's parser is tedious.
@1083706640 <cpage_!~Chris@charger.intuit.com> I put in a format() to see what's going on, and, sure enough, it's parsing the trailing comment as part of the character literal, e.g. in "myEnum = 'a' /* ' */" the literal is parsed as "'a' /* '". Whee!
@1083707015 <cpage_!~Chris@charger.intuit.com> Say, is there documentation for the regexp library?
@1083707502 <cpage_!~Chris@charger.intuit.com> ...on the website. Doesn't matter, the source code more or less says, "look at Perl."
@1083707593 -niven.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083710021 -niven.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1083710264 -niven.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083710351 [housel!housel@cantor.gwydiondylan.org] ACTION notes bhoult's stealth FD open-source announcement on c.l.l
@1083710635 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Hmm. Melange uses this pattern for character literals: '(\\\\?.)+'
@1083710642 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "'(\\\\?.)+'"
@1083710681 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Apparently, it's greedy, and so matches all interior apostrophes on a line.
@1083710699 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> It would make much more sense to match [^'] instead of ..
@1083710731 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It also needs to consume escaped apostrophes.
@1083710759 <housel!housel@cantor.gwydiondylan.org> yes
@1083710777 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm open to suggestions for how to best correct the expression.
@1083710991 <housel!housel@cantor.gwydiondylan.org> From the C lexer I'm using to test the lexer-generator in Monday's simple-parser library:
@1083710992 <housel!housel@cantor.gwydiondylan.org>       name c-char
@1083710992 <housel!housel@cantor.gwydiondylan.org>         = "[^'\\\\\n]|\\\\(['\"?\\abfnrtv]|[0-7]([0-7][0-7]?)?|x[0-9a-fA-F]+)";
@1083710992 <housel!housel@cantor.gwydiondylan.org>       token CHARACTER-CONSTANT = "L?'{c-char}+'",
@1083710992 <housel!housel@cantor.gwydiondylan.org>         semantic-value-function: extract-action;
@1083711043 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Maybe this?: '((\\')|[^'])+'
@1083711175 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, I think we're talking about the same pattern, if you simplify that.
@1083711175 <housel!housel@cantor.gwydiondylan.org> still incomplete... mine should handle everything except \u and \U
@1083711197 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This is within the context of a larger pattern that handles the other literals, too.
@1083711255 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To clarify: The Melange pattern for character literals only needs to identify the start/stop of the literal. Other code picks it apart.
@1083711280 <housel!housel@cantor.gwydiondylan.org> (the same applies to mine)
@1083711419 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I guess what I mean is, it looks like yours actually describes the entire syntax for character literals so a lexer can parse it with high fidelity, whereas Melange just uses regexp to find the boundaries, then it parses the interior separately.
@1083711489 <housel!housel@cantor.gwydiondylan.org> yes
@1083711584 <housel!housel@cantor.gwydiondylan.org> BTW, I just submitted bug 7000 as a tracking bug for the 2.3.12 release
@1083711607 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Interestingly, Melange's pattern for strings is: "(([^\\"]|\\.)*)"
@1083711636 <housel!housel@cantor.gwydiondylan.org> If there are any bugs that anyone thinks needs to be fixed before we release, please mark them as blocking bug 7000.
@1083711663 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So I guess whoever extended the character literal pattern just didn't notice the string pattern.
@1083711700 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Extended the character literal pattern to handle multiple characters. There's a note in the sources about handling it as a non-standard C extension.)
@1083711806 <housel!housel@cantor.gwydiondylan.org> The grammar in the C standard allows multiple characters, but the interpretation implementation-dependent
@1083711826 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Interesting. I didn't know that.
@1083711936 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm going to try a pattern that mimics the string pattern: '(([^\\']|\\.)+)'
@1083711965 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083712193 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, I had a typo and omitted a closing paren, which, of course, I didn't find out about until the code ran.
@1083712219 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This reminds me that some have asked me whether a Dylan compiler checks format strings at compile time.
@1083712258 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Of course, you could use something more like C++'s stream syntax (<<) to allow the compiler to parse and handle type-checking, but the code can be verbose.
@1083712348 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I wonder whether there could be a language feature where functions that accept strings that adhere to some DSL could describe the syntax to the compiler in the function definition.
@1083712377 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> http://www.common-lisp.net/paste/display/{537,539,542}
@1083712429 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (I don't like format strings, by the way, in case you couldn't tell from those...)
@1083712443 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The HTTP server complains about that URL.
@1083712488 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This works: http://www.common-lisp.net/paste/display/537,539,542
@1083712524 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Um.
@1083712540 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> The {x,y,z} notation means 'look at .../x, .../y, .../z', not that exact URL.
@1083712556 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh!
@1083712565 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> You're treating me like curl. :)
@1083712616 <housel!housel@cantor.gwydiondylan.org> no, like bash or csh
@1083712676 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "no" or "also"?
@1083712739 [housel!housel@cantor.gwydiondylan.org] ACTION admits ignorance wrt Curl
@1083712750 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In that case, it's "also". :)
@1083712756 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> No, curl, not Curl.
@1083712776 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Welcome to Pedants Anonymous.
@1083712781 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> My name is Chris.
@1083712784 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Curl is a web programming language (with an _EXTREMELY_ lame payment scheme) and the company that owns it.
@1083712795 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I have been pedantry-free for about five seconds.
@1083712833 <housel!housel@cantor.gwydiondylan.org> I'm still refusing to admit that I have a problem
@1083712838 <bhoult!~bhoult@morannon.eservglobal.co.nz> activity!
@1083712846 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Anyhow, have you people looked at my argument against format strings?
@1083712854 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Riastradh: So, what are your thoughts wrt those Lisp snippets?
@1083712854 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (partially mine and partially chandler's)
@1083712857 <bhoult!~bhoult@morannon.eservglobal.co.nz> nope
@1083712861 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Do so, then!
@1083712867 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Certainly the format string version is more compact.
@1083712877 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> cpage, can you understand what it does?
@1083712889 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I couldn't.  chandler had to make the second paste, the really verbose, vanilla Scheme one.
@1083712895 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ...for me to understand it, I mean.
@1083712896 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I can't understand what the verbose version does without a lot of reading, either.
@1083712926 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> What about the last one?
@1083712931 <bhoult!~bhoult@morannon.eservglobal.co.nz> so the basic argument is that -- like FORTRAN's format specifiers -- you can go too far and have an entire language within the format string?
@1083712965 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> bhoult, the argument is against those entire languages, not that you can have them.
@1083712998 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (Read SRFI 42 if you're interested in seeing how MYTHICAL-FORMAT's iteraction constructs work, by the way.)
@1083713039 <housel!housel@cantor.gwydiondylan.org> Dylan format strings are currently less powerful than printf
@1083713065 <housel!housel@cantor.gwydiondylan.org> a lot of code seems to avoid using them
@1083713067 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ...I see.
@1083713077 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I suppose the last one (542) begins to strike a balance between "all in a string" and "all in code".
@1083713085 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> cpage, yes.
@1083713094 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It's more like C++ stream syntax than the others.
@1083713105 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> It's supposed to be understandable without being as verbose as the Scheme one.
@1083713134 <bhoult!~bhoult@morannon.eservglobal.co.nz> as someone who codes in C++ all day, I'd usually far rather use printf than cout
@1083713143 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. even with the type-unsafeness
@1083713145 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Not really.  C++'s stream syntax is very mild syntactic sugar with no sophisticated _formatting_.
@1083713162 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> bhoult, what about MYTHICAL-FORMAT?
@1083713194 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. because it is SUCH a PITA to type all the quotes and << to replace a simple format string with a few inserts
@1083713200 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The real issue is that a static, linear text presentation doesn't fully meet the needs of the reader.
@1083713220 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I must depart now.  I'll read comments later.
@1083713240 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sometimes, a format string will be clearest. It probably depends on how much literal text you have vs. how many formatting directives.
@1083713340 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> One thing I'd appreciate, as I look at these regexp's that contain escaped characters that are further escaped when placed in Dylan string literals, is a way to specify a different escape character or different characters for the start/end of the string, so the interior can be simplified.
@1083713410 <housel!housel@cantor.gwydiondylan.org> yes... Lua uses % as its regexp escape character for that reason
@1083713471 <housel!housel@cantor.gwydiondylan.org> Maybe the regexp library functions could take an escape-character: keyword
@1083713478 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, one way to approach this would be to have a way to select a different escape character for the regexp library.
@1083713481 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Jynx!
@1083713850 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm getting some runtime exception messages complaining about the type of some argument, but it doesn't say which one. Is there a way to find out which argument is the problem?
@1083714786 <housel!housel@cantor.gwydiondylan.org> other than running in gdb and looking at the backtrace? no
@1083715120 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Why not use a macrological regexp package, like a Dylanized SRE facility?
@1083715221 <housel!housel@cantor.gwydiondylan.org> I've thought about that but couldn't come up with a syntax I liked
@1083715252 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Fine, then: revert Dylan to S-expressions and use SREs!
@1083715280 <housel!housel@cantor.gwydiondylan.org> I didn't say I liked SREs
@1083715292 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ...what don't you like about them?
@1083715371 <housel!housel@cantor.gwydiondylan.org> not as clear and compact as traditional regexp syntax
@1083715434 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I agree on the compactness, but I find that issue hardly very important; I vehemently disagree on the clarity, especially regarding the problems you were just discussing.
@1083715473 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I find it important to be able to express simple patterns compactly.
@1083715499 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If the verbose version is much longer than the string, it can be harder for the human to verify correctness.
@1083715500 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Simple patterns, fine: use POSIX regexps.  But for anything more complicated POSIX regexps get to be a complete mess.
@1083715524 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> If the string version is much less expressive than the SRE, it can be harder for the human to verify correctness.
@1083715527 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The trick is to come up with a system that allows both humans and compilers to verify correctness.
@1083715554 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 'Less expressive' wasn't the phrase I wanted, but I can't think of it right now...
@1083715576 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What is "SRE", btw?
@1083715586 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Symbolic Regular Expression.
@1083715606 <housel!housel@cantor.gwydiondylan.org> invented by Olin Shivers for scheme shell
@1083716345 <lisppaste!~lisppaste@69.55.225.112> housel pasted "My C lexer" at http://www.common-lisp.net/paste/display/876
@1083718326 -niven.freenode.net- buggs^z (~noidentd@pD9E9752C.dip.t-dialin.net) joined the channel
@1083726078 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Anybody have a reference for SRE? I can't find anything definitive in Google.
@1083726130 <housel!housel@cantor.gwydiondylan.org> http://www.ai.mit.edu/~shivers/sre.txt
@1083726204 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> http://www.scsh.net/docu/html/man-Z-H-8.html or so.
@1083726227 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 7, rather, not 8.
@1083728823 -dircproxy- You disconnected
@1083734679 -dircproxy- You connected
@1083737644 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1083737915 <bhoult!~bhoult@morannon.eservglobal.co.nz> hi andreas
@1083737930 <andreas!andreas@cantor.gwydiondylan.org> HI Bruce!
@1083743210 -niven.freenode.net- brucehoult (~bruce@202-0-45-94.adsl.paradise.net.nz) joined the channel
@1083744054 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning
@1083745234 <andreas!andreas@cantor.gwydiondylan.org> Morning Hannes.
@1083745678 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi andreas
@1083749772 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Quick poll: Is 0 a positive integer or merely non-negative?
@1083749800 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> F, T
@1083749898 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, do both positive? and negative? return false for 0?
@1083749923 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> i'd hope so
@1083749929 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And what about even? and odd? ?  The DRM doesn't cover these details, unfortunately.
@1083749961 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> I'd certain;y expect 0 to be even
@1083749978 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> anyway, it's very easy for you to find out what Gwydion does for all these cases
@1083749992 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sure, but I'm wondering what the answers *should* be.
@1083750018 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm wondering whether the DRM should have these points clarified.
@1083750068 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> it wouldn't hurt
@1083750114 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> define inline sealed method even? (a :: <integer>) => res :: <boolean>;
@1083750118 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz>   zero?(logand(a, 1));
@1083750118 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> end;
@1083750185 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, I'm reading about SRE's. I agree it would be good to have a Dylan implementation.
@1083750199 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Not sure what the syntax would be, though.
@1083750293 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> define sealed inline method positive? (num :: <real>) => res :: <boolean>;
@1083750297 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz>   num > 0;
@1083750297 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> end;
@1083750297 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> define sealed inline method negative? (num :: <real>) => res :: <boolean>;
@1083750297 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz>   num < 0;
@1083750297 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> end;
@1083750332 [brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz] ACTION would have been shocked had it been otherwise
@1083750351 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> this from: http://gwydiondylan.org/cgi-bin/viewcvs.cgi/src/d2c/runtime/dylan/num.dylan?rev=1.8.2.4&content-type=text/vnd.viewcvs-markup
@1083750978 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Thanks.
@1083751919 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Another detail I'm reminded of from looking at those definitions: The DRM doesn't say what happens if you call the logical bitwise operators with no arguments.
@1083751967 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> hmm
@1083751978 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> I thought they were defined to take exactly two (unlike in CL)
@1083751982 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> am I mistaken?
@1083752034 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> No, they take #rest. The source file you pointed at above shows implementation functions binary-foo() that take two args and and logfoo() uses reduce() on #rest.
@1083752201 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The DRM does say they take "zero or more...", which leaves me to conclude they should return zero, though an examination of the implementation of logand() shows it should return -1.
@1083752202 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> ah
@1083752231 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> i'd expect OR to return 0 and AND to return all 1's
@1083752255 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> it's the only logical result for zero arguments
@1083752275 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'd expect AND to return all 0's, because there are no 1's for it to AND with.
@1083752288 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> bad logic :-)
@1083752321 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Not that I think it matters too much, except that this should be specified in the DRM for completeness, or the functions should take two required args in addition to #rest.
@1083752323 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> bitand(a, b, c, d) = bitand( bitand(a, b, c), d)
@1083752375 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> reduce that a few times and you get bitand(bitand(), a) in the middle
@1083752381 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> .. whch must return a
@1083752406 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> ... which means bitand() must be all 1's
@1083752435 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> sorry, they're called logand.  YTWIM
@1083752461 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> er, YKWIM
@1083752478 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I don't see how that necessarily follows. In fact, speaking abstractly, why doesn't logand() return no value or the empty list?
@1083752491 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yes, IKWYM :-)
@1083752523 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> isn't the return type defined as <integer>?
@1083752550 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yes. I was "speaking abstractly".
@1083752576 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Not that I'm suggesting it should do either of those things, I'm just saying I don't necessarily see a clear reason for saying it should return all 1's.
@1083752576 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> personally, i'd probably prefer it was defined to take (a, #rest b), not (#rest b)
@1083752592 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why not (a, b, #rest r) ?
@1083752595 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> and then the question wouldn't arise
@1083752613 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why leave open the question of what the default value is when only one or zero args are supplied?
@1083752625 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> that would be OK, too, but it might force some special cases in some code
@1083752630 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, I guess we mean the same thing.
@1083752691 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, the math operators are strictly binary.
@1083752748 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'd kinda like to see a consistency here. Either they all take #rest or two args, etc., or they all have two flavors, one with two args and one with #rest.
@1083752827 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, ideally I'd like to see them all take just #rest and have some guarantee that the compiler will always optimize the common case of two args.
@1083752832 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> hard to make infix operators take a #rest
@1083752900 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Are they infix? Isn't it just the expression syntax that's infix? The functions themselves don't have to be binary, do they/
@1083752901 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ?
@1083752918 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> so you could call \+(1, 2, 3, 4)  ??
@1083752922 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> maybe so
@1083752967 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I just don't see any difference between \+ and \logior in that regard.
@1083753018 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...except that Dylan is defined to be efficient rather then philosophically pure, so it makes sense to have two required args for the common math operators.
@1083753086 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...unless, of course, you could figure out a way to make sure the compiler would always produce the same code even if you defined them to take #rest instead.
@1083753122 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> One thing I've been mulling over is proposing support for methods that take varying numbers of required args.
@1083753148 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That could help with efficiency in these cases. You could define the generic with #rest, but have methods with two required args for the common cases.
@1083753257 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> There may or may not be any real advantage, though. I have to think about it some more.
@1083753292 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The basic idea comes from overloading in C++. It's sometimes useful to be able to differentiate methods based upon the number of args supplied.
@1083753303 -niven.freenode.net- pritlove (~pritlove@194.95.203.217) joined the channel
@1083753384 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> brucehoult: Actually, it just dawned on my that you're right about logand(), but I'm seeing it from another angle: logand(a) returns a, implying that it's ANDed with all 1's.
@1083753444 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...so logand() must return all 1's.
@1083753815 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> that's right
@1083753859 [brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz] ACTION to bed
@1083753886 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> g'night!
@1083753912 <brucehoult!~bruce@202-0-45-94.adsl.paradise.net.nz> and unless I'm mistaken it 0400 there!
@1083753928 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Almost.
@1083755345 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083756286 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083759061 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083759820 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083766743 <cgay!~cgay@ita4fw1.itasoftware.com> minimal-console-compiler leaves a lot to be desired in terms of error reporting...it stuffs them away into log files and doesn't tell you where they are.
@1083766780 <cgay!~cgay@ita4fw1.itasoftware.com> I vaguely remember there was a script to grovel through the logs and find them, but I guess it should just display them directly.
@1083766836 <housel!housel@cantor.gwydiondylan.org> the errors ought to go to *standard-output* as well as the log
@1083766851 <cgay!~cgay@ita4fw1.itasoftware.com> yeah
@1083766951 <cgay!~cgay@ita4fw1.itasoftware.com> in fact, i can't even find the "3 serious warnings" in the logs...
@1083766979 <housel!housel@cantor.gwydiondylan.org> they should be in library-name-build/library-name.log
@1083767004 <housel!housel@cantor.gwydiondylan.org> (or Build/library-name/library-name.log, depending)
@1083767079 <cgay!~cgay@ita4fw1.itasoftware.com> yeah, they aren't there
@1083767153 <housel!housel@cantor.gwydiondylan.org> hmmm.... could that be what the release notes (and bug 6981) were referring to?
@1083767181 <cgay!~cgay@ita4fw1.itasoftware.com> i'm probably screwing up somehow.  i'll poke around more
@1083767236 <cgay!~cgay@ita4fw1.itasoftware.com> hmm.  the koala-build dir has today's date, but all the files in it haven't changed since Mar 20
@1083767278 <housel!housel@cantor.gwydiondylan.org> did you start using personal-root?  in which case there should be a Build dir under that
@1083767356 <cgay!~cgay@ita4fw1.itasoftware.com> yes.  i'm building koala-app and the output for koala-app itself is in <registry>/Build/koala-app
@1083767366 <cgay!~cgay@ita4fw1.itasoftware.com> but no warnings in there about the subprojects
@1083767403 <housel!housel@cantor.gwydiondylan.org> could it be the 'build' vs. 'Build' problem I wonder?
@1083767418 <housel!housel@cantor.gwydiondylan.org> need to add that to bugzilla...
@1083767425 <cgay!~cgay@ita4fw1.itasoftware.com> bingo!
@1083767445 <cgay!~cgay@ita4fw1.itasoftware.com> i hadn't typed 'g' in my dired, so i didn't see the 'build' dir
@1083767496 <cgay!~cgay@ita4fw1.itasoftware.com> so it seems to use 'build' for the subprojects and 'Build' for the top-level library
@1083767509 <housel!housel@cantor.gwydiondylan.org> bletch
@1083767805 <cgay!~cgay@ita4fw1.itasoftware.com> there aren't many occurrances of 'Build'.  i'm tempted to just go change them all and try rebuilding fd.
@1083767836 <cgay!~cgay@ita4fw1.itasoftware.com> at any rate, i'll bugzilla it
@1083767928 <cgay!~cgay@ita4fw1.itasoftware.com> i wish bugzilla wouldn't ellipsize summaries :(
@1083769061 <cgay!~cgay@ita4fw1.itasoftware.com> i think fd should make it much easier to specify loose/tight mode when building, both in the IDE and in the command line
@1083769143 <housel!housel@cantor.gwydiondylan.org> how would you change it? build-tight and build-loose, "Build interactive mode" menu item etc?
@1083769164 <cgay!~cgay@ita4fw1.itasoftware.com> one ought to be able to do 'build -tight myapp" just before doing a checkin, for example
@1083769182 <cgay!~cgay@ita4fw1.itasoftware.com> (i just found a bug that would have been caught that way)
@1083769211 <cgay!~cgay@ita4fw1.itasoftware.com> in the IDE you have to go to Project -> Settings and change it.  and it's not clear how that affects subprojects either
@1083769440 <housel!housel@cantor.gwydiondylan.org> (I just think the Settings tab is way too verbose... tooltips would be a better way of expressing all the effects of those options)
@1083769495 <cgay!~cgay@ita4fw1.itasoftware.com> in the IDE there's a "Build..." (or something like that) menu item that puts up a dialog first.  that would be a good place to do it.  just a radio button for tight and loose, and a checkbox that says 'for subprojects too'
@1083769853 <housel!housel@cantor.gwydiondylan.org> "Advanced Build..." I think
@1083769888 <cgay!~cgay@ita4fw1.itasoftware.com> yuk
@1083770211 <housel!housel@cantor.gwydiondylan.org> do you know if the IDE designers used HCI consultants or did user walkthroughs?
@1083770290 <cgay!~cgay@ita4fw1.itasoftware.com> i think there was some of that.  cpage would know.
@1083770782 -niven.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1083771839 -dircproxy- You disconnected
@1083772403 -dircproxy- You connected
@1083772448 -dircproxy- You disconnected
@1083773515 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1083776233 -niven.freenode.net- pritlove (~pritlove@port-212-202-201-16.dynamic.qsc.de) joined the channel
@1083781929 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1083782015 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm happy to say we did do a round of user-testing. We used a company in MA and I flew out to attend. Andy, Scott, Hotchkiss, and myself were there (I don't recall if there was anyone else).
@1083782174 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> We got some valuable feedback. Unfortunately, we only did one round, we later added some things, and the tests were not as detailed as we'd have liked.
@1083782355 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> On the one hand, we really were concerned with providing a good user experience, but on the other, we had time constraints and needed to ship something.
@1083782449 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Another issue is that several people contributed to the UI and, not to insult anyone, but not everyone was as experienced with doing UI.
@1083782576 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Also, when you have several people with strong opinions (and I mean that in a good way), you don't always get complete consensus on every detail.
@1083782613 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [That's my way of saying: If there's anything about the UI you like, I did it, but if you don't like it, it's somebody else's fault. :-]
@1083782849 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, Hqn Dylan was my first real Windows app, and I must say I found several ways in which Windows put constraints on what we could do, forcing less than desirable UI in some cases. For example, I did the search feature, modeling it on CodeWarrior. It uses the Shift key to change commands between forward and backwards search. Unfortunately, Windows doesn't support dynamic menu items that change as you toggle modifier keys, so 
@1083782849 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> all the menu items had to be in the menu at all times, doubling the number of items.
@1083782898 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> CodeWarrior for Windows has the same problem.
@1083782915 <cgay!~cgay@ita4fw1.itasoftware.com> that is good for discoverability though
@1083782977 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sure, which is why I insisted on putting them in the menu instead of going the other way and invisibly handling the shift key.
@1083783108 <cgay!~cgay@ita4fw1.itasoftware.com> i'm just saying i wouldn't consider that a problem for Windows...more like a problem for the Mac (i assume it's the Mac)
@1083783109 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Now that we have the sources I'd love to revisit the UI, especially any decisions I made that I may now disagree with.
@1083783226 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I don't know if it's a problem for new users, but this Shift convention started probably ten or more years ago, so every experienced Mac programmer is certainly aware of it.
@1083783271 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...and nowadays, using Shift to reverse search direction is in the Apple HI guidelines and appears in every Cocoa app by default.
@1083783328 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Of course, the standard Find menu only contains a few items, so they put the "Search Backwards" command in the menu.
@1083783404 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> CodeWarrior has enough menu items it's important to be able to simplify the menu. Perhaps they could put just "Search Backwards" in the menu to aid in discoverability of the other reversed commands.
@1083783493 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I actually considered that for HD, but this isn't/wasn't a standard convention for Windows apps (most of the search commands and associated keys are non-standard for Windows), so I felt it was important for that audience to put them all in the menu.
@1083783597 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, the last time I looked at FunDev I came to the conclusion that Windows simply looks ugly and forces bad UI overall, no matter how good your designs. :-(
@1083783636 <cgay!~cgay@ita4fw1.itasoftware.com> yeah, i can understand it in the Mac, where it's been the convention since the dawn of time.
@1083783694 <cgay!~cgay@ita4fw1.itasoftware.com> i consider the recent 'innovation' in Windows menus, where they remove items that haven't been used for a long time, to be horrible UI.  i instantly turn it off whenever i get a new Windows install.
@1083783704 <cgay!~cgay@ita4fw1.itasoftware.com> except you can't turn it off everywhere :(
@1083783706 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> +1
@1083783713 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> They remove items that haven't been used in a long time!!!?
@1083783719 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Some items.
@1083783727 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> That's _horrible_!
@1083783737 <cgay!~cgay@ita4fw1.itasoftware.com> then if you leave the mouse there for a while, like you can't find something, they expand the menu to show all items
@1083783745 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> They're placed in a section at the bottom of the menu represented by a single item. You click it and the other items appear below.
@1083783745 <cgay!~cgay@ita4fw1.itasoftware.com> icky icky icky patoo!
@1083783776 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The problem, of course, is apps with too many menu items.
@1083783803 <cgay!~cgay@ita4fw1.itasoftware.com> MS Word comes to mind.  (i haven't found a way to turn that feature off in it)
@1083783830 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Einstein's quote regarding geniuses and making things simple comes to mind...there aren't enough geniuses to keep apps simple enough.
@1083784010 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> On the other hand, some people on more than one occasion argued against putting everything in menu items, simply to not clutter the menus, which isn't the right answer, either.
@1083784041 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It certainly takes a lot of time and effort to make good UI.
@1083784118 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...rather: argued against making everything discoverable in the UI (which most often means "in a menu item", but not always).
@1083784199 -niven.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1083784202 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> heya
@1083784397 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [Looking at streams.dylan] I've come across a place where the Mindy version of some code uses several methods with singleton specializers, but there's one method for d2c, which uses a select statement, with the following comment:
@1083784406 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> / The compiler can't deal with singleton(<byte>).
@1083784433 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is that still true? Should this code be revisited, or is it still necessary for bootstrapping?
@1083784442 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://gwydiondylan.org/cgi-bin/viewcvs.cgi/src/common/streams/streams.dylan?rev=1.8&content-type=text/vnd.viewcvs-markup>
@1083784492 -dircproxy- You connected
@1083784496 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (it's the implementation of type-for-sequence())
@1083784828 <housel!housel@cantor.gwydiondylan.org> It's still true
@1083784842 <housel!housel@cantor.gwydiondylan.org> ... and I'll explain why after I eat lunch
@1083784853 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Excellent. :)
@1083784953 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083784955 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1083787321 <housel!housel@cantor.gwydiondylan.org> ok, here's the deal...
@1083787402 <housel!housel@cantor.gwydiondylan.org> to specialize on singleton(something) the compiler needs to be able to guarantee at compile-time that something is \== to an equivalent something-else
@1083787426 <housel!housel@cantor.gwydiondylan.org> define constant <byte> = limited(<integer>, min: 0, max: 255);
@1083787458 <housel!housel@cantor.gwydiondylan.org> right now d2c doesn't guarantee that any two values returned by limited(<integer>, min: 0, max: 255) will be \== to each other
@1083787511 <housel!housel@cantor.gwydiondylan.org> that would require memoization tables at both compile-time and run-time
@1083787543 <housel!housel@cantor.gwydiondylan.org> ... something we're not doing yet
@1083787611 -dircproxy- You disconnected
@1083787861 -dircproxy- You connected
@1083789015 -niven.freenode.net- hannes (~hannes@pD9E7BD85.dip.t-dialin.net) joined the channel
@1083789016 <hannes!~hannes@pD9E7BD85.dip.t-dialin.net> re
@1083789174 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1083789180 <hannes!~hannes@pD9E7BD85.dip.t-dialin.net> hi peter
@1083789801 <bhoult!~bhoult@morannon.eservglobal.co.nz> morning!
@1083790008 <housel!housel@cantor.gwydiondylan.org> hi bruce
@1083790163 <bhoult!~bhoult@morannon.eservglobal.co.nz> the MPW cmd-F G H and shift for backwards seems to have been widely copied.  F= find dialog box, G = find again, H = find the selected text
@1083790229 <bhoult!~bhoult@morannon.eservglobal.co.nz> The SYmantec products (THINK C/Pascal) used cmd-E for "enter the selection as the search string, but don't actually look for it yet", and that seems to have often been picked up as a (simultaneous) alternative to MPWs cmd-H
@1083791142 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> housel: I don't understand how that impacts the code in question, though. If you define a constant <byte> and use it everywhere, it'll always be ==, won't it?
@1083791281 <bhoult!~bhoult@morannon.eservglobal.co.nz> no, it's the *type* that isn't always ==
@1083791289 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right.
@1083791300 <bhoult!~bhoult@morannon.eservglobal.co.nz> let <a> = limited(<integer>);
@1083791309 <bhoult!~bhoult@morannon.eservglobal.co.nz> let <b> = limited(<integer>);
@1083791327 <bhoult!~bhoult@morannon.eservglobal.co.nz> a and b should (perhaps) be ==
@1083791327 <housel!housel@cantor.gwydiondylan.org> if you do it, it will work, but the compiler will give you a warning every time you use a library with that method
@1083791339 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If you use "limited(<integer>, min: 0, max: 255)" more than once, they may not be ==, but if you assign it to a constant <byte> and use "<byte>" everywhere it should be okay, right?
@1083791350 <housel!housel@cantor.gwydiondylan.org> that's currently a minor wart in newio
@1083791379 <bhoult!~bhoult@morannon.eservglobal.co.nz> cpage: I think so
@1083791406 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh! The problem is the type singleton(<byte>).
@1083791420 <bhoult!~bhoult@morannon.eservglobal.co.nz> actually, I'd prefer it that way.  There used to be disagreement between Pascal implementors about whether composite types had name equivilence or structure equivilence
@1083791476 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...or is it?
@1083791487 <housel!housel@cantor.gwydiondylan.org> d2c tends to take the tack that functional objects are to be made \== whenever possible
@1083791516 <housel!housel@cantor.gwydiondylan.org> so fixing this would be a good thing
@1083791523 <bhoult!~bhoult@morannon.eservglobal.co.nz> name equivilence means that you can go "type personName = char[20], address = char[20];" and the type types aren't assignment compatable
@1083791574 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think perhaps name equivalence only makes sense for static languages.
@1083791600 <bhoult!~bhoult@morannon.eservglobal.co.nz> that's possible, though I don't see it
@1083791628 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Perhaps I have it reversed, because the word "equivalence" threw me off.
@1083791701 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Two instantiations with the same arguments to a type constructor may not produce the same value in a dynamic language, but in a statically-typed language there is no such thing as runtime instances of types.
@1083791748 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, anyway, does "singleton(<foo>)" memoize?
@1083791787 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Do "f(x == <byte>)" and "g(x == <byte>)" have the same signature?
@1083791862 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...er...actually, those are unrelated questions, I guess.
@1083791866 <bhoult!~bhoult@morannon.eservglobal.co.nz> if you replaced <byte> with <integer> or <float> or almost anything else , then, yes
@1083791927 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I still don't understand why it's a problem with <byte>.
@1083791953 <housel!housel@cantor.gwydiondylan.org> because the compiler doesn't think it knows the address of that <limited-type> instance
@1083792001 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Even if it gets the type from a constant?
@1083792029 <housel!housel@cantor.gwydiondylan.org> yes
@1083792035 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why?
@1083792049 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Pardon me if I seem dense.)
@1083792053 <housel!housel@cantor.gwydiondylan.org> define constant $my-table = make(<object-table>);
@1083792062 <housel!housel@cantor.gwydiondylan.org> it doesn't know the address of $my-table either
@1083792082 <housel!housel@cantor.gwydiondylan.org> because make(<object-table>) isn't compile-time evaluable
@1083792136 <housel!housel@cantor.gwydiondylan.org> it's the same priniciple as far as d2c is concerned
@1083792205 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What does d2c do with type specifiers that are not compile-time evaluable?
@1083792232 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (I must admit I'm not sure what the DRM says about this.)
@1083792239 <housel!housel@cantor.gwydiondylan.org> it evaluates them at run-time
@1083792259 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Okay, so there's an efficiency issue, but is there a semantic problem?
@1083792287 <housel!housel@cantor.gwydiondylan.org> no, as I said, the code does in fact work
@1083792296 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, I didn't notice you'd said that.
@1083792316 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That explains my apparent density. :)
@1083792411 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, do I understand correctly that the issue is that \select is faster than method dispatch for runtime types?
@1083792465 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...that == isn't a problem, since \select is using it for the comparison.
@1083792486 <housel!housel@cantor.gwydiondylan.org> hard to say without timing it
@1083792532 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Okay, so, I'm sorry if this sounds like we're going in circles, but: Why does this version of type-for-file-stream() exist?
@1083792670 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...and perhaps more relevantly, type-for-sequence(), which uses singleton(<byte>).
@1083792693 <housel!housel@cantor.gwydiondylan.org> you'd have to ask Ben Folk-Williams or whoever wrote that code to be sure
@1083792727 <housel!housel@cantor.gwydiondylan.org> I think I did make some improvements in the behaviour of specializing on singletons of limited types last fall, but I can't remember now
@1083792735 <housel!housel@cantor.gwydiondylan.org> so things might have been worse then
@1083792762 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> AH.
@1083792769 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/H/h
@1083793048 <housel!housel@cantor.gwydiondylan.org> hmmm... looks like what I fixed was <byte-character>, which is an internal magic type
@1083799490 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083799648 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1083800317 -dircproxy- You disconnected
@1083800666 -dircproxy- You connected
@1083803346 -dircproxy- You disconnected
@1083803708 -dircproxy- You connected
@1083804570 <housel!housel@cantor.gwydiondylan.org> according to "svn blame", Ben Folk-Williams is indeed responsible for the #if(mindy) in question
@1083804749 -niven.freenode.net- buggs|afk (~noidentd@pD9E95F69.dip.t-dialin.net) joined the channel
@1083806530 -dircproxy- You disconnected
@1083806847 -niven.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1083806899 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> housel: are you here?
@1083808720 -dircproxy- You connected
@1083808774 <housel!housel@cantor.gwydiondylan.org> I am now
@1083811662 -niven.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1083816276 -dircproxy- You disconnected
@1083820149 -dircproxy- You connected
@1083822421 <CIA-2!micah@hyperreal.info> commit by housel to src: Bug: 7002
@1083822421 <CIA-2!micah@hyperreal.info> Start of work to make bootstrapping (both mindy and d2c) possible.
@1083822714 <CIA-2!micah@hyperreal.info> commit by housel to src: Bug: 7002
@1083822715 <CIA-2!micah@hyperreal.info> Merge in changes made between Linux Alpha 3 and the full
@1083822715 <CIA-2!micah@hyperreal.info> open-source release.
@1083823013 <CIA-2!micah@hyperreal.info> commit by housel to src: Bug: 7002
@1083823013 <CIA-2!micah@hyperreal.info> Start of work to make bootstrapping (both mindy and d2c) possible.
@1083823426 <CIA-2!micah@hyperreal.info> commit by housel to src: Bug: 7002
@1083823426 <CIA-2!micah@hyperreal.info> Merge in changes made between Linux Alpha 3 and the full
@1083823426 <CIA-2!micah@hyperreal.info> open-source release.
@1083830633 <^self!~fn@66.209.66.105> so, you have the tree imported into svn?
@1083836822 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1083838541 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> hi chris.  just getting up?  or still up?
@1083838840 -niven.freenode.net- brucehoult (~bruce@202-0-60-199.adsl.paradise.net.nz) joined the channel
@1083839759 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Still up. Hoping to go to bed soon.
@1083839797 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Discovering one more problem with Melange's parser each time I fix one.
@1083839878 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The latest problem is it chokes on lines split with a '\'. It isn't ignoring the backslash like it should and it throws off the token lexer.
@1083840073 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> fun fun fun
@1083840095 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i woke up at 4am for some reason and couldn't sleep, so i'm hacking on koala
@1083840279 <brucehoult!~bruce@202-0-60-199.adsl.paradise.net.nz> well, it might only be 10:45 here, but I think I'm off to bed anyway...
@1083840924 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1083841455 -niven.freenode.net- OneSadCookie (~keith@203-79-121-211.cable.paradise.net.nz) joined the channel
@1083841470 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> any melange gurus here?
@1083841537 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> anybody at all here?
@1083841803 [hannes!~hannes@port-195-158-171-122.dynamic.qsc.de] ACTION is here but is no melange guru
@1083841865 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> :)
@1083842083 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> same here
@1083842458 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> anyway, not a melange problem at the moment
@1083842466 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> d2c doesn't like my 4000-line exports file :(
@1083842510 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> it segfaults almost immediately with 15000 stack frames
@1083842614 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> I thought it was due to too many things exported from one export statement
@1083842624 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> but reducing it to one item per statement doesn't change anything
@1083843036 <andreas!andreas@cantor.gwydiondylan.org> Hi!
@1083843056 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> howdy
@1083843063 <andreas!andreas@cantor.gwydiondylan.org> OneSadCookie: Macro expansion is handled recursively, with one recursion per entry in the statement.
@1083843069 <andreas!andreas@cantor.gwydiondylan.org> Play with ulimit -s.
@1083843088 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> so that applies at the define module level?
@1083843105 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> 'cos putting only one exported name into each export statement had no effect
@1083843124 <andreas!andreas@cantor.gwydiondylan.org> Yes, it's about expanding the "define module" macro.
@1083843156 [OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz] ACTION does ulimit -s 65536
@1083843166 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> yay, no segfaul
@1083843168 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> +t
@1083843196 <andreas!andreas@cantor.gwydiondylan.org> One day, we will have guard pages and proper exceptions on stack overrun. :)
@1083843274 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> I'd prefer it just handled it iteratively :p
@1083843313 <andreas!andreas@cantor.gwydiondylan.org> You're welcome to submit an implementation. :)
@1083843363 <OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz> thanks :)
@1083843376 [OneSadCookie!~keith@203-79-121-211.cable.paradise.net.nz] ACTION puts on to-do list for 2059
@1083851614 <CIA-2!micah@hyperreal.info> commit by cgay to libraries: job: minor
@1083851617 <CIA-2!micah@hyperreal.info> Clean up http error response codes a bit.  This way it's clear at the call site
@1083851617 <CIA-2!micah@hyperreal.info> what the arguments to the error function are.
@1083851868 <CIA-2!micah@hyperreal.info> commit by cgay to libraries: job: minor
@1083851868 <CIA-2!micah@hyperreal.info> Use system:operating-system:load-library instead of
@1083851868 <CIA-2!micah@hyperreal.info> win32-kernel:win32-kernel:LoadLibrary.  Comment out body of unload-module
@1083851868 <CIA-2!micah@hyperreal.info> because there isn't a corresponding unload-library function in the os
@1083851868 <CIA-2!micah@hyperreal.info> module yet (and unload-module really isn't needed now anyway).
@1083853722 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: btw, i think #key headers is not used (and will not be used) in http-error definer macro
@1083853741 <cgay!~cgay@ita4fw1.itasoftware.com> yes, it's used in one place
@1083853801 -niven.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083853802 <cgay!~cgay@ita4fw1.itasoftware.com> in static-files.dylan: not-modified(headers: ...)
@1083853803 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1083853871 <cgay!~cgay@ita4fw1.itasoftware.com> i hope i didn't just break something with that checkin...i'm getting this weird error now:
@1083853883 <cgay!~cgay@ita4fw1.itasoftware.com> [bin] mars| ./koala-app
@1083853883 <cgay!~cgay@ita4fw1.itasoftware.com> Cannot add-method {<getter-method>: ??? (<avalue>)} in {<library>} to sealed generic {<sealed-generic-function>: avalue-alist}.
@1083853883 <cgay!~cgay@ita4fw1.itasoftware.com> zsh: 17458 trace trap  ./koala-app
@1083853883 <cgay!~cgay@ita4fw1.itasoftware.com> exit status: 133
@1083853910 <cgay!~cgay@ita4fw1.itasoftware.com> but i did see it before, in windows.  come to think of it, i think it is related to compiling in tight mode.
@1083854012 [hannes!~hannes@port-195-158-171-122.dynamic.qsc.de] ACTION hit ctrl + alt + backspace accidentaly :(
@1083854043 <cgay!~cgay@ita4fw1.itasoftware.com> does that restart X ?  (vague memories...)
@1083854052 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> it quits X
@1083854082 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> i wanted to hit ctrl + alt + enter for a new xterm ;)
@1083854136 <cgay!~cgay@ita4fw1.itasoftware.com> yikes.  you might want to consider changing some key bindings
@1083854227 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> he, alt + enter also opens a new xterm... so i already changed key bindings, but still used the old binding ;)
@1083854274 <housel!housel@cantor.gwydiondylan.org> it seems to be a frequent prank to ask people to do a backward-kill-sexp in emacs (M-C-Backspace)
@1083854474 -dircproxy- You disconnected
@1083854507 -dircproxy- You connected
@1083854896 -dircproxy- You disconnected
@1083855214 -dircproxy- You connected
@1083855396 -dircproxy- You disconnected
@1083855415 -dircproxy- You connected
@1083858038 -niven.freenode.net- gabor (root@cantor.gwydiondylan.org) joined the channel
@1083858063 <gabor!root@cantor.gwydiondylan.org> re.
@1083858108 <housel!housel@cantor.gwydiondylan.org> hi gabor
@1083858119 <gabor!root@cantor.gwydiondylan.org> hi peter!
@1083858800 -dircproxy- You disconnected
@1083862343 -niven.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1083862704 -dircproxy- You connected
@1083863470 -niven.freenode.net- hannes (~hannes@217.231.182.22) joined the channel
@1083863471 <hannes!~hannes@217.231.182.22> re
@1083864047 <^self!~fn@66.209.66.105> hi
@1083866179 <andreas!andreas@cantor.gwydiondylan.org> Hi!
@1083866257 <housel!housel@cantor.gwydiondylan.org> I've just been trying out the new fsfs backend of subversion on the gd repository
@1083866264 <housel!housel@cantor.gwydiondylan.org> It's a lot more space-efficient:
@1083866274 <housel!housel@cantor.gwydiondylan.org> 225012  gd/db
@1083866274 <housel!housel@cantor.gwydiondylan.org> 556286  gd-bdb/db
@1083866326 <housel!housel@cantor.gwydiondylan.org> 398886	/var/lib/cvs
@1083866644 <housel!housel@cantor.gwydiondylan.org> (actually the CVS figure is an underestimate, because the subversion databases are the result of merging the old CMU RCS repository and the current CVS one)
@1083868378 -niven.freenode.net- pritlove (~pritlove@d213-102-40-241.cust.tele2.ch) joined the channel
@1083868627 <hannes!~hannes@217.231.182.22> hi tim
@1083872410 -niven.freenode.net- bfulgham (~chatzilla@63.192.83.15) joined the channel
@1083875353 <housel!housel@cantor.gwydiondylan.org> hi brent
@1083875353 <^self!~fn@66.209.66.105> eh.
@1083875790 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1083875816 <andreas!andreas@cantor.gwydiondylan.org> With the new backend, I'm inclined to reconsider subversion.
@1083875816 <housel!housel@cantor.gwydiondylan.org> hi andreas
@1083875856 <housel!housel@cantor.gwydiondylan.org> yes... I think we should switch over when 1.1 is released
@1083876443 -niven.freenode.net- pritlove (~pritlove@10.24.76.83.cust.bluewin.ch) joined the channel
@1083876828 -niven.freenode.net- jemfinch|lambda (~jfincher@dhcp065-024-075-056.columbus.rr.com) joined the channel
@1083876855 <jemfinch|lambda!~jfincher@dhcp065-024-075-056.columbus.rr.com> where can I find more information on this statement about Dylan: "Exceptions are object-based. Handlers can resume or abort the code  that signaled the exception. The exception system is built on a  documented substrate for non-local flow of control."
@1083876894 <housel!housel@cantor.gwydiondylan.org> you could read the Dylan Reference Manual, or you could ask us...
@1083876973 <jemfinch|lambda!~jfincher@dhcp065-024-075-056.columbus.rr.com> how about I do both? :)  Where's the Dylan Reference Manual?
@1083876976 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/books/drm/drm_52.html is the chapter on the exception system
@1083881346 -niven.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) left the channel
@1083881393 -niven.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) joined the channel
@1083883027 -niven.freenode.net- buggs^z (~noidentd@pD9E95F69.dip.t-dialin.net) joined the channel
@1083885657 -niven.freenode.net- pritlove (~pritlove@152.191.62.81.cust.bluewin.ch) joined the channel
@1083893302 -dircproxy- You disconnected
@1083893320 -kornbluth.freenode.net- You joined the channel
@1083894491 -kornbluth.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083894562 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is there a writeup for the 2000 ICFP? We're talking about it over in a private chat and I wanted to relay the details of what happened WRT OpenGL. I know some of you have been working on improved OpenGL support since then.
@1083895245 -kornbluth.freenode.net- pritlove (~pritlove@152.191.62.81.cust.bluewin.ch) joined the channel
@1083898265 <bhoult!~bhoult@morannon.eservglobal.co.nz> No, I don't think we have a writeup for 2000
@1083898284 <bhoult!~bhoult@morannon.eservglobal.co.nz> I manage to get motivated to write one when we're hours from getting slashdotted
@1083898295 <bhoult!~bhoult@morannon.eservglobal.co.nz> no one else has yet been motivated to write one
@1083898322 <bhoult!~bhoult@morannon.eservglobal.co.nz> slashdotting didn't happen in 2000 or 2002 because we didn't do that well
@1083898975 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Can you remind me what the issue was? I thought there was some problem where it couldn't be done (completely?) in Dylan.
@1083899865 <bhoult!~bhoult@morannon.eservglobal.co.nz> problem?
@1083899866 <bhoult!~bhoult@morannon.eservglobal.co.nz> no
@1083899868 <bhoult!~bhoult@morannon.eservglobal.co.nz> in 2000?
@1083899889 <bhoult!~bhoult@morannon.eservglobal.co.nz> we just didn't get all the features finished in time
@1083899960 <bhoult!~bhoult@morannon.eservglobal.co.nz> http://www.cis.upenn.edu/proj/plclub/contest/faq.php
@1083899981 <bhoult!~bhoult@morannon.eservglobal.co.nz> The ICFP contest is again (like last year) going to be ab offline one, with just a text result file submitted
@1083899999 <bhoult!~bhoult@morannon.eservglobal.co.nz> that may mean that safety again doesn't matter :-(
@1083900012 <bhoult!~bhoult@morannon.eservglobal.co.nz> It may mean that people with lots of CPUs have an advantage
@1083900041 <bhoult!~bhoult@morannon.eservglobal.co.nz> I just had a really evil idea, and Keither an Alex and I are following up with our various contacts at the moment.
@1083900064 <bhoult!~bhoult@morannon.eservglobal.co.nz> Clue: http://news.bbc.co.uk/2/hi/technology/3672887.stm
@1083905928 -dircproxy- You connected
@1083906218 <housel!housel@cantor.gwydiondylan.org> hmmm
@1083906266 <bhoult!~bhoult@morannon.eservglobal.co.nz> ?
@1083906347 <housel!housel@cantor.gwydiondylan.org> your clue... it has a ring of plausibility
@1083906384 <bhoult!~bhoult@morannon.eservglobal.co.nz> I'd hate for ICFP to just be an arms race
@1083906397 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. but if it is, why take a knife to a gunfight?
@1083906607 <bhoult!~bhoult@morannon.eservglobal.co.nz> It would make for a good /. story too :-)
@1083906659 <bhoult!~bhoult@morannon.eservglobal.co.nz> anyway, we're asking.  THe worst that can happen is they say "no"
@1083909776 -kornbluth.freenode.net- brucehoult (~bruce@202-0-61-111.adsl.paradise.net.nz) joined the channel
@1083912268 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Bug: 7003
@1083912268 <CIA-2!micah@hyperreal.info> Until the old build-system is eliminated, document the confusing array
@1083912268 <CIA-2!micah@hyperreal.info> of substitution variables it uses.
@1083920665 <gabor!root@cantor.gwydiondylan.org> bruce: what OS is that renderfarm running?
@1083920704 [gabor!root@cantor.gwydiondylan.org] ACTION thinks "IRIX" is the wrong answer
@1083920764 <gabor!root@cantor.gwydiondylan.org> isn't there an XGrid for trial somewhere?
@1083920970 -kornbluth.freenode.net- pritlove (~pritlove@56.254.186.195.cust.bluewin.ch) joined the channel
@1083920981 <gabor!root@cantor.gwydiondylan.org> hi tim
@1083923954 -kornbluth.freenode.net- brucehoult (~bruce@202-0-60-209.adsl.paradise.net.nz) joined the channel
@1083924148 <brucehoult!~bruce@202-0-60-209.adsl.paradise.net.nz> gabor: linux I think
@1083924406 <gabor!root@cantor.gwydiondylan.org> hi bruce! x86 linux?
@1083924567 <brucehoult!~bruce@202-0-60-209.adsl.paradise.net.nz> I don't knot
@1083924568 <brucehoult!~bruce@202-0-60-209.adsl.paradise.net.nz> know
@1083924588 <brucehoult!~bruce@202-0-60-209.adsl.paradise.net.nz> may be ppc
@1083924675 <brucehoult!~bruce@202-0-60-209.adsl.paradise.net.nz> it's a long shot anyway
@1083926293 -kornbluth.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1083926295 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1083927479 -kornbluth.freenode.net- jemfinch|lambda (~jfincher@dhcp065-024-075-056.columbus.rr.com) joined the channel
@1083927499 <gabor!root@cantor.gwydiondylan.org> hi hannes, jem
@1083940565 -kornbluth.freenode.net- jemfinch|lambda (~jfincher@140.254.57.33) joined the channel
@1083940665 <jemfinch|lambda!~jfincher@140.254.57.33> hmm.
@1083941257 -kornbluth.freenode.net- hannes (~hannes@pD9E7B830.dip.t-dialin.net) joined the channel
@1083941519 <gabor!root@cantor.gwydiondylan.org> hmmm?
@1083941540 <housel!housel@cantor.gwydiondylan.org> hmmm.
@1083941632 <jemfinch|lambda!~jfincher@140.254.57.33> I was wondering how easy it would be to implement Python in Dylan.
@1083941637 <hannes!~hannes@pD9E7B830.dip.t-dialin.net> re
@1083941637 <jemfinch|lambda!~jfincher@140.254.57.33> and how fast the resulting interpreter would go.
@1083942488 <gabor!root@cantor.gwydiondylan.org> just try it. I once wrote a mc68k assembler interpreter in dylan, it was interesting
@1083942524 <jemfinch|lambda!~jfincher@140.254.57.33> gabor: meh, I'm probably not good enough, especially since I don't know Dylan right now.
@1083942539 <jemfinch|lambda!~jfincher@140.254.57.33> gabor: Dylan can do dynamic classes, though, right?
@1083942594 <gabor!root@cantor.gwydiondylan.org> not d2c. FD can. But you probably won't map python classes to dylan classes anyway
@1083942705 <jemfinch|lambda!~jfincher@140.254.57.33> why not?
@1083942721 <jemfinch|lambda!~jfincher@140.254.57.33> Dylan can do metaclasses, can't it?
@1083942947 <gabor!root@cantor.gwydiondylan.org> because a pyton class may better be modelled as a <table> in dylan. Esp. in an interpreter.
@1083942988 <gabor!root@cantor.gwydiondylan.org> no, dylan has only one metaclass: <class>, and since it has no MOP, you cannot create more
@1083943014 <gabor!root@cantor.gwydiondylan.org> see the FAQ in wiki: monday.sf.net/wiki
@1083943036 <jemfinch|lambda!~jfincher@140.254.57.33> why does Dylan have no MOP?
@1083943143 <gabor!root@cantor.gwydiondylan.org> it is hidden in the compiler. you open a can of worms (performancewise, not conceptually) when you make the MOP accessible to users
@1083943173 <jemfinch|lambda!~jfincher@140.254.57.33> yeah.
@1083943194 <jemfinch|lambda!~jfincher@140.254.57.33> I don't know if I'd make the same trade-off if I were writing my own language, though.
@1083943293 <jemfinch|lambda!~jfincher@140.254.57.33> anyway, I gotta run :)
@1083943298 <jemfinch|lambda!~jfincher@140.254.57.33> thanks for the help.
@1083943299 -kornbluth.freenode.net- jemfinch|lambda (~jfincher@140.254.57.33) left the channel
@1083944343 -kornbluth.freenode.net- pritlove (~pritlove@81.218.186.195.cust.bluewin.ch) joined the channel
@1083946938 -dircproxy- You disconnected
@1083948535 -dircproxy- You connected
@1083950404 -kornbluth.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1083951252 -kornbluth.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1083951253 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1083953482 -kornbluth.freenode.net- buggs^z (~noidentd@pD9E97D1E.dip.t-dialin.net) joined the channel
@1083954757 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1083954790 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sometimes it seems like you can't win. Either somebody complains Dylan isn't dynamic and introspective enough, or it isn't "static enough."
@1083954822 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Just yesterday someone was arguing that Dylan allowed "accidental dynamism
@1083954828 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [ignore that]
@1083954859 <housel!housel@cantor.gwydiondylan.org> accidental dynamism?
@1083954894 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Just yesterday someone was arguing that Dylan allowed "accidental dynamism," in the sense that if there was a function named, say, "open()," and you defined your own method with the same name, you might unintentionally introduce dispatching overhead at runtime.
@1083954911 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I said that was simply not likely to occur and was never a problem in real-world programming.
@1083954948 <housel!housel@cantor.gwydiondylan.org> that's what fine-grained control over namespaces is for
@1083954991 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, well, then he complained when I said a real-world program would probably have open-file() and open-window(), not just an open() function for both. He balked that meant you had to put "warts" on names.
@1083955012 <cgay!~cgay@ita4fw1.itasoftware.com> bah
@1083955050 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I argued that a certain amount of disambiguation makes code more readable, and that "creating a file descriptor" and "displaying a UI window" are two separate concepts, anyway.
@1083955075 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Anyway, can you see the stuff I have to deal with? Feh! :-)
@1083955087 <cgay!~cgay@ita4fw1.itasoftware.com> where was that discussion taking place?
@1083955114 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Private chat.
@1083955151 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To be fair, I remember thinking something similar when someone explained generic functions to me back in 1989.
@1083955248 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But what's really going on here is he's coming from the C++ world where you really care about micro-performance and you only use "virtual" when you're sure you need it. There's a certain level of paranoia about performance, which I can understand, but switching to Dylan means having to become comfortable with a slightly different way to look at programming.
@1083955345 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I also appreciate the desire to be able to use variables to indicate namespaces so you could write my-window.open() or some-file.open() as in C++. However, that's just not compatible with the idea of dynamic typing and optional type declarations.
@1083955414 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The real trick is that it's necessary to get someone to commit to learn enough Dylan that they can look at real-world programs to become familiar enough with Dylan to decide to pursue it further.
@1083955439 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...for many people, at least.
@1083955594 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To do that, you often need a "killer app." Either a library that does something they need enough to warrant writing application code in Dylan, or a program they want to use and modify. But everybody wants something different. So when I have these discussions about Dylan I get a million different "does it have X" questions, which can sometimes be a little overwhelming.
@1083955612 <housel!housel@cantor.gwydiondylan.org> yes
@1083955719 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> People are simply skeptical when I claim Dylan is better, and when I try to show small examples I may get some "oh, yes, I see" responses, but I also often get "but that's just one small thing." The killer feature of Dylan is its sum total. There isn't really one feature you can point at and have most people go, "oh, yes, that's what I need."
@1083955795 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So the trick is getting someone to listen long enough to see the overall picture of Dylan. Maybe I should host a Dylan class and give away a free lunch with it.
@1083955818 <hannes!~hannes@pD9E7B830.dip.t-dialin.net> btw, where are docs about the new svn backend andreas mentioned yesterday?
@1083955823 <housel!housel@cantor.gwydiondylan.org> right... it was built by people with experience using dynamic programming languages, and knowing what worked, not as a one-trick pony
@1083955850 <housel!housel@cantor.gwydiondylan.org> hannes: just as sec, let me find the URL
@1083955939 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To be more accurate, there are several people who keep coming back for more discussions and to ask more questions, but most of them have yet to actually try the language.
@1083955952 <cgay!~cgay@ita4fw1.itasoftware.com> why has NFS been causing X applications to hang since 1991 or earlier?  argh.
@1083955969 <housel!housel@cantor.gwydiondylan.org> http://web.mit.edu/ghudson/info/fsfs
@1083956573 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The open-window, open-file, open-fire example is informative. I remember asking how to get a ~delete-style destructor on Dylan objects so I could deallocate window resources. Then I realised that hiding the lifecycle with what is, in effect, a big fat virtual method called delete(), could be Considered Harmful. Clear, explicit naming is a win over ambiguous, overly-overloaded short names.
@1083956628 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I haven't been a grad student, but I still find this funny: <http://www.phdcomics.com/comics/archive.php?comicid=396>
@1083956788 <housel!housel@cantor.gwydiondylan.org> http://www.phdcomics.com/comics/archive.php?comicid=296 is probably my favorite
@1083956806 <housel!housel@cantor.gwydiondylan.org> that's about where I am in my Ph.D. program right now...
@1083956930 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, in the specific case of delete, I think that may not be an issue. New and delete are so ubiquitous it's pretty clear what's going on. For comparison, most of the standard Dylan names are short. This is fine when the concepts are very generic and commonly used, like make(). In fact, I've always been a little bothered by object-class(). Why isn't it just class()?
@1083957029 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To answer my own question, although "foo.class" might seem nicer than "foo.object-class", the fact is you shouldn't be looking directly at classes very often (instead, you should use instance?() or a purpose-built predicate for your class).
@1083957076 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> housel: heh
@1083957176 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But delete's ubiquity makes it tempting to use it to hide lifecycle rather than just as a failsafe clean-up mechanism. Which bites you when you move from C++ to Java...
@1083957299 <housel!housel@cantor.gwydiondylan.org> with-open-frobozz()...end; is a much nicer way of managing lifetimes
@1083957317 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm not sure how delete hides lifetime.
@1083957336 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yes, I generally prefer with- macros to make the scope/lifetime more visible.
@1083957342 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I've argued that, myself.
@1083957367 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think delete actually makes the lifetime more visible than using stack allocation.
@1083957378 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Yes, I like with....
@1083957442 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I agree with others who complain that Java doesn't have destructors, because it also doesn't have with- macros.
@1083957451 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think you really must have one or the other.
@1083957464 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But delete is the very, very end of the lifecycle. It's more for the compiler or malloc(). You want close-window(), shut-file(), cease-fire() long before then. And in Java, the finalizer may not get called for ages, or at all, so the paradigm breaks there, (Erm, Java has finalise(), but it won't get called until GC, or perhaps not at all).
@1083957471 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Requiring everyone to write explicit cleanup clauses is just asking for bugs.
@1083957517 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> with-* is better than delete(). Scope guards in C++ are the equivalent of with-*.
@1083957522 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right, finalization isn't lifetime management.
@1083957547 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But delete gets used as both finalization & lifecycle...
@1083957647 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The problem with { MyType x; } is that it isn't clear there's a life-cycle really going on there. In real-world code where you have a bunch of locals, it isn't clear which ones are "just variables" and which ones are "executing functions at start/end of lifetime." Perhaps this is also more of a concern in a language that leans towards side-effecting code.
@1083957700 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...especially when you consider { MyType x, *y; }. x is automatically deleted, y isn't.
@1083957770 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I use WithWhatever whateverGuard(whatever); 
@1083957774 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> robmyers: I'm still not clear on how delete hides life-cycle, though. It's clear that delete ends the life of an object, just as it's clear that new (or make) is the beginning.
@1083957785 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Which looks like a with-macro()
@1083957800 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> It doesn't neccessarily hide it, but misuse by programmers means that it does.
@1083957809 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yes, that's closer to with-.
@1083957829 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The classic example being the C++ programmer who converts to Java and writes a database class that disposes of the connection in finalize()...
@1083957973 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> On a separate subject, have you seen Moon's comments on s-expresions being a bad representation for programs (at the data level, not syntactically):  http://lispmeister.com/cgi-bin/blosxom/lisp-news/moon-on-s-expressions.html 
@1083958008 <housel!housel@cantor.gwydiondylan.org> yes I saw that (on cll yesterday)
@1083958034 <housel!housel@cantor.gwydiondylan.org> I agree completely, though I think a lot of c.l.l people don't get it
@1083958037 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> There's a whole page of comments on Paul Graham's site. I'm intrigued by the idea that Moon worked on a Dylan compiler after leaving Apple.
@1083958046 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Really? That's interesting. I'll have to read that.
@1083958053 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Well they're so used to bracket trolls... :-)
@1083958117 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I thought a lot of the problems could be worked around by fatteningf up symbols to allow more data to be attached.
@1083958164 <housel!housel@cantor.gwydiondylan.org> with proplists, symbols are already fat enough
@1083958261 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> True. So you could add the source to the proplist, maybe...
@1083958276 <housel!housel@cantor.gwydiondylan.org> conses don't have proplists, though
@1083958386 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But a funtion name symbol would, wouldn't it?
@1083958408 <housel!housel@cantor.gwydiondylan.org> the main pain is all of the unnecessary destructuring you have to do over and over again
@1083958441 <housel!housel@cantor.gwydiondylan.org> it's better to build a parse tree that has exactly the fields you need in its AST nodes
@1083958475 <housel!housel@cantor.gwydiondylan.org> and where you can distinguish between different constructs by looking at the AST node type
@1083958509 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Maybe someday I can explain to you the representation of programs I used 
@1083958509 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> in the Dylan compiler I wrote (never finished) after I left Apple."
@1083958518 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I wonder what he's referring to.
@1083958528 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> ME tOO
@1083958538 <housel!housel@cantor.gwydiondylan.org> he says later he'd have to fetch it from backup
@1083958542 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Liam, get off the caps lock0...
@1083958576 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But he wrote that ages ago, it's from the ARC suggestions page. Maybe we can gently remind him. :-)
@1083958594 <housel!housel@cantor.gwydiondylan.org> 2 years ago I think
@1083958748 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> So CLOS would be better for writing a Lisp compiler. But that's still built on Lisp.
@1083958762 <housel!housel@cantor.gwydiondylan.org> sure
@1083958825 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right. A compiler and compiled code are abstractions and should have the insulation of an interface.
@1083958967 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hmmm. You could still represent the code as mutable data through an interface.
@1083959376 [housel!housel@cantor.gwydiondylan.org] ACTION wishes he knew where fundev/Sources/lib/run-time/pentium-win32/make-reloc.exe came from
@1083959609 <hannes!~hannes@pD9E7B830.dip.t-dialin.net> housel: from /pizza/mushroom/noer/beta17/build/i386-cygwin32-x-i386-cygwin32/newlib/libc/sys/cygwin32/ ;)
@1083959703 <housel!housel@cantor.gwydiondylan.org> heh
@1083960383 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> cpage: Regarding new IDEs - http://acg.media.mit.edu/people/jarfish/codefocus/
@1083960967 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Thanks. I've seen those before. I especially love Plate.
@1083961018 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Prolog is interesting. It's s-expressions under the hood (well, clauses), remapped by its operator system. so foo([1,2,3], X):- X = 1. becomes something like :-(foo, .(1 .(2) .(3)), X, =(X, 1)). Arc wants something similar for its "syntax". 
@1083961029 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm anxious to see more of their work.
@1083961064 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Now, Dylan could be represented clausally (everything's a function call), and \+ . \:= etc. could be remapped by an operator system, just like Prolog. It would make macros easier. :-)
@1083961075 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (When I wrote that I meant codefocus, but I'm also interested in Arc.)
@1083961093 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> cpage: Yes, they haven't done much recently (both cases. :-) )
@1083961138 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> whoops, .(1 .(2 .(3))), well, you get the idea...
@1083961315 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> While I'm too worn out at this point in the week to have an intense argument, I think _Moon_ doesn't get it, not Lispers.  S-expressions aren't intended to be the actual AST that the compiler operates on; I don't know of any compilers that use S-expressions as the actual AST representation, except for Twobit (where S-expressions are just a convenient representation when it was written; the AST code could be rewritten to use it
@1083961363 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION tries to imagine Moon not getting Lisp. ;-)
@1083961454 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Moon's point isn't about what compilers use, it's about what program-writing programs use.
@1083961467 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> He's saying that cons cells aren't abstract enough.
@1083961483 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> With which I agree.
@1083961522 <housel!housel@cantor.gwydiondylan.org> I think Scheme is a lot better in this regard, because Scheme makes an explicit program/data distinction
@1083961528 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> How much more abstract can you get? A single cell? :-)
@1083961534 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> An explicit program/data distinction?
@1083961589 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Cons cells are implementation details. You don't usually get much lower-level than a single cell.
@1083961590 <housel!housel@cantor.gwydiondylan.org> if you look at RnRS, it never says that load calls read or anything like that
@1083961618 <housel!housel@cantor.gwydiondylan.org> the semantics are defined in terms of program syntax, not in terms of symbols and cons cells
@1083961641 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Nor is LOAD particularly useful in arguments such as these, as it is well-known that it is both optional and a bad design.
@1083961722 <housel!housel@cantor.gwydiondylan.org> plus, as Moon notes, hygienic macros require something more
@1083961823 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Certainly; I'm not disputing that.  But that's not an argument against using S-expressions for syntax.  Macro implementations, and compilers, are free to choose whatever AST representation they like.
@1083961856 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Moon explicitly says he's talking about compiling, not syntax.
@1083961900 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Then he's not really arguing anything useful, is he?  As I said, rarely do any Lisp compilers really use S-expressions for the AST.
@1083962075 -kornbluth.freenode.net- brucehoult (~bruce@202-0-60-62.adsl.paradise.net.nz) joined the channel
@1083962164 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> No, I think the point is what program-writing programs use, and that includes macros.
@1083963713 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hmmm... How do I access a class slot in CLOS without having an instance of the class?
@1083963830 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I assume it's the same as in Dylan: You don't.
@1083963864 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Ew.
@1083963898 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Class allocation is an implementation detail you aren't supposed to be concerned with.
@1083963917 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> OK, I'll move the instances hashtable to a global, then. :-)
@1083963936 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If an interface wants to give you access to something not requiring an object, it exports a function that doesn't require an object, along with the class and its slot accessors.
@1083963977 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Speaking from a Dylan perspective. I can't speak authoritatively about CLOS.)
@1083964168 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION makes a mental note of the pattern.
@1083964260 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> If you want class slots -- and not each subclass slots --, just use a global variable.
@1083964270 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In contrast, in C++ or Java almost everything is inside one class declaration or another. In Dylan (and I presume CLOS) class definitions are much simpler entities and you use a combination of pieces to define an interface, orthogonally to class definitions.
@1083964314 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Oh, yes, classes aren't namespaces in Dylan/CLOS. They are in C++/Java. I often find myself torn between a namespace and a class in C++...
@1083964387 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Scott Meyers essentially recommends leaning the Dylan way. Keep your class declarations simpler and use non-member functions when possible.
@1083965800 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Goodnight, all.
@1083971599 -kornbluth.freenode.net- pritlove (~pritlove@26.152.62.81.cust.bluewin.ch) joined the channel
@1083975750 -kornbluth.freenode.net- hannes (~hannes@217.231.162.242) joined the channel
@1083975752 <hannes!~hannes@217.231.162.242> re
@1083978920 -dircproxy- You disconnected
@1083982479 -kornbluth.freenode.net- pritlove (~pritlove@26.152.62.81.cust.bluewin.ch) joined the channel
@1083982479 -kornbluth.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083982479 -kornbluth.freenode.net- chandler (~chandler@64-145-60-36.client.dsl.net) joined the channel
@1083982582 -kornbluth.freenode.net- pritlove (~pritlove@26.152.62.81.cust.bluewin.ch) joined the channel
@1083982582 -kornbluth.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1083982582 -kornbluth.freenode.net- chandler (~chandler@64-145-60-36.client.dsl.net) joined the channel
@1083985097 -dircproxy- You connected
@1083985522 -dircproxy- You disconnected
@1083986607 -dircproxy- You connected
@1083986617 <housel!housel@cantor.gwydiondylan.org> re
@1083987073 -dircproxy- You disconnected
@1083987188 -dircproxy- You connected
@1083989169 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Job: minor
@1083989171 <CIA-2!micah@hyperreal.info> Remove a patch artifact.
@1083990014 -dircproxy- You disconnected
@1083992997 -dircproxy- You connected
@1083995276 -kornbluth.freenode.net- brucehoult (~bruce@202-0-45-13.adsl.paradise.net.nz) joined the channel
@1083997932 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What's a "Job," anyway? I've seen "Job: foo" in some commit comments.
@1083998363 <brucehoult!~bruce@202-0-45-13.adsl.paradise.net.nz> it's just alternative syntax to "Bug:"
@1083998385 <brucehoult!~bruce@202-0-45-13.adsl.paradise.net.nz> and there are some fixed names you can use instead of bug humbers
@1083998428 <brucehoult!~bruce@202-0-45-13.adsl.paradise.net.nz> http://gwydiondylan.org/cgi-bin/cvszilla/job.cgi
@1083998493 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So if I correct spelling errors on a web page I don't have to open a bug? I can just commit the change with "Job: website"?
@1083998556 <housel!housel@cantor.gwydiondylan.org> yes
@1083998569 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Excellent.
@1083998602 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, the link on that page that says it goes to a description of each job is b0rken.
@1083998671 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The "click here" link.
@1083998902 <housel!housel@cantor.gwydiondylan.org> BTW unless there are objections I'm going to commit bug 7004 (removing fundev/build) tomorrow morning (PDT)
@1083998930 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The build directory got into the repository?
@1083999029 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [Looks at the bug.] OIC.
@1084010150 <gabor!root@cantor.gwydiondylan.org> cpage: re. object-class, you cannot have class here because it is a define-word
@1084011616 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought define words were only significant after "define". No?
@1084011646 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In fact, the name of the macro is "class-definer".
@1084012634 -kornbluth.freenode.net- pritlove (~pritlove@26.152.62.81.cust.bluewin.ch) joined the channel
@1084013018 -dircproxy- You disconnected
@1084013540 <gabor!root@cantor.gwydiondylan.org> yes, but class-definer macro reserves a define-word "class" (IIRC)
@1084013565 <gabor!root@cantor.gwydiondylan.org> not that sure anymore ;-)
@1084014085 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> yes, but aren't define words only significant within a "define"?
@1084016063 [gabor!root@cantor.gwydiondylan.org] ACTION does not know
@1084016090 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm pretty sure "define words" don't exist.
@1084016119 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm pretty sure "define" takes the define word, adds the "-definer" suffix, then resolves the name to get the appropriate macro.
@1084016176 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ie., "class" is not a name in the dylan module.
@1084016233 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> However, from the human perspective, it may make more sense to give the function object-class() this longer name, because it should rarely be used.
@1084016360 <gabor!root@cantor.gwydiondylan.org> the BNF specifies define-body-word and define-list-word
@1084016422 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sure, the compiler stores the names in a table, but they aren't in the module namespace. They're looked up explicitly within certain contexts. (AFAIAA.)
@1084016572 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This is surprising: I was looking through the DRM index for "define word" when I came across "define inert domain". The page it points to correctly calls it "define sealed domain". Somebody forgot to delete the old "inert" index entry.
@1084016752 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_79.html#HEADING79-0>
@1084016772 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "The name bound to a definition macro is the macro's define-body-word or define-list-word suffixed by "-definer". This suffixing convention is analogous to the naming convention for setters and allows the define-body-word or define-list-word to be used for another purpose."
@1084016809 <gabor!root@cantor.gwydiondylan.org> inert ---> typo this is in the errata
@1084016855 <gabor!root@cantor.gwydiondylan.org> good reference, btw.
@1084016890 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In the BNF, define-body-word is only in the grammar for "define ..." and "end ...".
@1084016895 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_118.html#HEADING118-24>
@1084016907 <gabor!root@cantor.gwydiondylan.org> anyway, overloading \class in a way you suggest is not healthy. I still prefer object-class.
@1084016974 <gabor!root@cantor.gwydiondylan.org> The following reserved words are exported by the Dylan module:
@1084016975 <gabor!root@cantor.gwydiondylan.org> begin-word:
@1084016976 <gabor!root@cantor.gwydiondylan.org> one of begin block case for if method
@1084016979 <gabor!root@cantor.gwydiondylan.org> one of select unless until while
@1084016980 <gabor!root@cantor.gwydiondylan.org> function-word:
@1084016981 <gabor!root@cantor.gwydiondylan.org> (none)
@1084016983 <gabor!root@cantor.gwydiondylan.org> define-body-word:
@1084016984 <gabor!root@cantor.gwydiondylan.org> one of class library method module
@1084016986 <gabor!root@cantor.gwydiondylan.org> define-list-word:
@1084016987 <gabor!root@cantor.gwydiondylan.org> one of constant variable domain
@1084017004 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I didn't "suggest" it, though, I merely discussed it.
@1084017005 <gabor!root@cantor.gwydiondylan.org> from http://www.gwydiondylan.org/drm/drm_117.htm#HEADING117-0
@1084017018 <gabor!root@cantor.gwydiondylan.org> ok :-)
@1084017129 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The core set: define end handler let local macro otherwise
@1084017140 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I love that that's smaller than the keywords list of C.
@1084017183 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 'Course, I think it would be even better if handler, let, local, and otherwise weren't core words.
@1084017513 <gabor!root@cantor.gwydiondylan.org> but we agree now that "class" is reserved?
@1084017628 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "reserved" in what sense? As far as I am aware, "class" only means something when in the define-boyd-word position in the grammar.
@1084017735 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> There is no module binding "class," and it is not a core word recognized specially by the compiler as is "end," for example.
@1084017752 <gabor!root@cantor.gwydiondylan.org> at least d2c handles "class" (as verbatim) specially as opposed to \class
@1084017784 <gabor!root@cantor.gwydiondylan.org> so you can write "let \class = 42" but not "let class = 42"
@1084017831 <gabor!root@cantor.gwydiondylan.org> they belong to different syntactic categories
@1084017855 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The DRM discussion of "reserved words" is actually referring only to the "core words" in the grammar.
@1084017902 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Are you saying that if I define a definer "define macro i ..." that I could no longer use "i" as a local variable name?
@1084018027 <gabor!root@cantor.gwydiondylan.org> basically yes. (for sure when it is a function macro, or a statement macro)
@1084018055 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, of course for function and statement macros. They define unadorned names.
@1084018069 [gabor!root@cantor.gwydiondylan.org] ACTION needs to check whether this is true for definition macros too
@1084018075 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But "define macro foo" defines "foo-definer", not "foo".
@1084018137 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_20.html#HEADING20-0>
@1084018200 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [Ignore that URL.
@1084018202 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ]
@1084018205 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_17.html#HEADING17-0>
@1084018219 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "A reserved word is a syntactic token that has the form of a name but is reserved by the Dylan language and so cannot be given a binding and cannot be used as a named value reference. There are seven reserved words in Dylan: define, end, handler, let, local, macro, and otherwise."
@1084018238 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_15.html#HEADING15-0>
@1084018293 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "A user-defined defining macro is a macro that defines a definition in terms of other constructs. A call to a user-defined defining macro always begins with the word define and includes the name of the defining macro. This name when suffixed by "-definer" is the name of a visible binding whose value is the defining macro."
@1084018351 <gabor!root@cantor.gwydiondylan.org>  // definition-syntax-info{<define-body-macro-definition>}
@1084018353 <gabor!root@cantor.gwydiondylan.org> //   -- method on imported GF
@1084018354 <gabor!root@cantor.gwydiondylan.org> //
@1084018356 <gabor!root@cantor.gwydiondylan.org> // For body-style define macros, strip off the -definer.
@1084018357 <gabor!root@cantor.gwydiondylan.org> // 
@1084018359 <gabor!root@cantor.gwydiondylan.org> define method definition-syntax-info
@1084018360 <gabor!root@cantor.gwydiondylan.org>     (defn :: <define-body-macro-definition>, name :: <symbol>)
@1084018362 <gabor!root@cantor.gwydiondylan.org>     => (real-name :: false-or(<symbol>), category :: <symbol>);
@1084018363 <gabor!root@cantor.gwydiondylan.org>   values(sans-definer(name), #"define-body");
@1084018365 <gabor!root@cantor.gwydiondylan.org> end method definition-syntax-info;
@1084018408 <gabor!root@cantor.gwydiondylan.org> this is from the CVS
@1084018412 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.gwydiondylan.org/books/drm/drm_78.html#HEADING78-0>
@1084018431 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "The new define-body-word or define-list-word becomes a partially reserved word in each module where the macro definition is visible. In particular a define-body-word or define-list-word cannot be used as a modifier in a definition. It can still be used as a variable-name."
@1084018482 <gabor!root@cantor.gwydiondylan.org> yes: It can still be used as a variable-name.
@1084018519 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> define variable class = 4;
@1084018525 [gabor!root@cantor.gwydiondylan.org] ACTION has to look whether d2c does the right thing
@1084018594 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I wonder whether the grammar parses that as "define {adjectives} class..." or "define variable {name}...".
@1084018654 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That quote seems to be saying you can't use "variable" as a modifier of "class," implying that it would be parsed as "define variable {name}..." but those could be orthogonal.
@1084018856 <gabor!root@cantor.gwydiondylan.org> define variable class = 0;
@1084018857 <gabor!root@cantor.gwydiondylan.org> begin
@1084018859 <gabor!root@cantor.gwydiondylan.org>         let class = 0;
@1084018860 <gabor!root@cantor.gwydiondylan.org> end;
@1084018861 <gabor!root@cantor.gwydiondylan.org> Both work in d2c :-)
@1084018877 <gabor!root@cantor.gwydiondylan.org> cpage: you win
@1084018900 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yay! :-)
@1084018944 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And that answers my question about how it parses "variable class".
@1084018965 <gabor!root@cantor.gwydiondylan.org> variable cannot be an adjective (it is partially reserved) so it is the second parse
@1084018988 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Phew! I was concerned it might mean define words collide with modifier words.
@1084019013 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, I got that wrong.
@1084019113 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> There *is* a collision between modifiers and define body words, but the result is that it handles the first word it finds that matches a define body word as a...define body word.
@1084019149 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...rather than meaning it would prohibit "variable" from appearing before the word "class".
@1084019193 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, now we can write:
@1084019210 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   define constant class = object-class;
@1084019231 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> and everybody who thinks "object-class" is too long are happy.
@1084019256 <gabor!root@cantor.gwydiondylan.org> right
@1084019270 <gabor!root@cantor.gwydiondylan.org> or just rename on import
@1084019303 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That's cheating. :-)
@1084019358 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Speaking of grammars and parsing, I found and fixed another bug in Melange.
@1084019476 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If I can get it to parse all of the Mac OS X Carbon headers, my next goal will be to have it handle umbrella frameworks automatically.
@1084019533 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> You currently have to list every single framework that is indirectly used, making for long command line option lists and massive wasted time finding out which frameworks are used.
@1084019550 <gabor!root@cantor.gwydiondylan.org> good idea
@1084019574 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I've been at it for days.
@1084019600 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Well, including the time spent fixing parser errors.)
@1084019675 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I was trying to build an interface for <Carbon/Carbon.h>, but it took forever to get to the errors.
@1084019722 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm now just working with <QD/QD.h>, but it surprisingly uses things like CFNetwork indirectly.
@1084019886 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Just added another --framework to my ever-longer command line. If this takes much longer I'm just going to implement umbrella support before proceeding.
@1084019906 -kornbluth.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084019908 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1084019912 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> howdy
@1084020489 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In fact, it isn't clear to me why you have to supply --framework to Melange at all.
@1084020518 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It already has a list of paths to search for frameworks, and the #include path tells it the name of the framework to use.
@1084025602 -dircproxy- You connected
@1084025806 -kornbluth.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084025812 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084025850 <housel!housel@cantor.gwydiondylan.org> hi rob
@1084026184 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Peter
@1084030765 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Bug: 7004
@1084030765 <CIA-2!micah@hyperreal.info> Remove these files; they are (obsolete) copies of
@1084030765 <CIA-2!micah@hyperreal.info> fundev/Sources/dfmc/c-run-time.
@1084032397 -kornbluth.freenode.net- pritlove (~pritlove@86.152.186.195.cust.bluewin.ch) joined the channel
@1084032864 -kornbluth.freenode.net- hannes (~hannes@pD9E7A2F2.dip.t-dialin.net) joined the channel
@1084032866 <hannes!~hannes@pD9E7A2F2.dip.t-dialin.net> re
@1084039894 -kornbluth.freenode.net- buggs^z (~noidentd@pD9E97F25.dip.t-dialin.net) joined the channel
@1084040249 -kornbluth.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1084040265 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> hi y'all
@1084041143 <housel!housel@cantor.gwydiondylan.org> hi rayiner
@1084042464 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> oh, design question: What's the most Dylan-esque way to parse an XML file?
@1084042523 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> In Python, I have the parsers be more or less auto-generated, using a lot of getattr/setattr's. Dylan doesn't have dynamic classes, though, IIRC
@1084042558 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Job: minor
@1084042558 <CIA-2!micah@hyperreal.info> Remove this sparc executable; gema is easy to build.
@1084042619 <housel!housel@cantor.gwydiondylan.org> I've designed a SAX-like interface based on protocols
@1084042657 <housel!housel@cantor.gwydiondylan.org> FunDev does support class creation at runtime, though I've never been in a situation where I wanted to use it
@1084042715 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> I didn't know fundev did that. I use it quite a lot in Python, but treating classes as tables is a Python idiom
@1084042920 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> its pretty useful for this particular sort of thing, though --- my parser for gcc-xml's output files is driven by a big description table auto-generated by a script that analyzes a bunch of sample files to determine their format
@1084043088 <housel!housel@cantor.gwydiondylan.org> You might want to look at fundev/Sources/web/expat
@1084044282 -kornbluth.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) left the channel
@1084048300 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Didn't Andreas write an XML system?
@1084049724 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Dylan supports class and type creation at runtime. AFAIR, FunDev is a complete implementation of Dylan. d2c, unfortunately, is missing this feature.
@1084049794 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I've always wondered whether creating classes at runtime was really something an application should do, or if that's just something useful for the development environment.
@1084049845 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ie., creating classes to represent your data at runtime seems a bit like self-modification, which can make programs harder to understand.
@1084052475 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The Fun-O documentation for "keyed-by" is terse. I can't figure out what it's supposed to do: <http://www.functionalobjects.com/products/doc/core/core_10.htm#HEADING10-0>
@1084053982 <housel!housel@cantor.gwydiondylan.org> what's so hard to understand about it?
@1084054102 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The description is completely opaque to me. What does "allows iteration over table elements" mean?
@1084054122 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What is the type of "my-key"? Is it a function?
@1084054244 <housel!housel@cantor.gwydiondylan.org> if you have a table with #"a" => 1, #"b" => 2, then for(value keyed-by key in table) ... end will execute once with value=1 key=#"a", once with value=2 key=#"b"
@1084054285 <housel!housel@cantor.gwydiondylan.org> it works for <sequence> too, in which case the key argument will be integers
@1084054346 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, so "keyed-by foo" introduces a binding "foo" and is assigned the key of the current element.
@1084054370 <housel!housel@cantor.gwydiondylan.org> yes
@1084054417 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, I found it misleading that the description only says "allows iteration over table elements," as if that's all it did.
@1084054442 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I also find the name "keyed-by" misleading, as it sounds to me like select's "by".
@1084054462 <housel!housel@cantor.gwydiondylan.org> what would you call it?
@1084054475 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I've been trying to decide.
@1084054507 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What about changing the syntax so it looks more like a multiple-value let?
@1084054526 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> for ((my-element, my-key) in my-collection)
@1084054559 <housel!housel@cantor.gwydiondylan.org> it's not clear in that case whether the key or the value should come first
@1084054565 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right.
@1084054586 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> for ((my-element, #key my-key) in my-collection)  :-)
@1084054677 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "key," "and-key"
@1084054777 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> for (e and-key k in my-collection)
@1084054835 <housel!housel@cantor.gwydiondylan.org> or with-key maybe
@1084054867 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Maybe, although that sounds a bit like the key is a filter or a parameter, when it's really a binding just like the element.
@1084054895 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ie., the problem I have with the syntax is that it isn't clear what's being bound and what's a "parameter" for the iteration.
@1084055077 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, if "using" is almost always used for the backward-iteration-protocol, I'd like to see "backward" supported to make the common case clearer.
@1084055096 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> for (e in my-sequence backward)
@1084055380 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In fact, the name "keyed-by" kinda makes me want:
@1084055382 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   for (e in c keyed-by k)
@1084055390 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> where k is a sequence of keys.
@1084055671 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...or a function that generates keys.
@1084055712 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   for (e each-key k in c)  for (e with-key k in c)  for (e and k in c)
@1084055804 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   for (e key: k in c)
@1084056296 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   for (key k and e in c)
@1084056432 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   for (key k element e in c)
@1084056617 [housel!housel@cantor.gwydiondylan.org] ACTION has been using it the way it is for 7 years with no complaints
@1084056640 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION hasn't been using it because it's so hard to fathom
@1084056645 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION :-P
@1084056697 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm certain it can be learned and used correctly without a huge effort, but as someone coming to it fresh I find it a little unobvious.
@1084056717 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If there was a clear winner it would be worth finding out what it was.
@1084056771 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It certainly sticks out a little compared to all the other words supported by "for". It's the only one that's hyphenated.
@1084056792 <housel!housel@cantor.gwydiondylan.org> When I first saw it I just remember thinking "oh, neat, they found a way of exposing some otherwise hard-to-access iterator-protocol functionality"
@1084056820 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, the documentation certainly didn't lead me to that reaction. Better documentation would help.
@1084056860 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But in the context of the language I really do find "keyed-by" leading me to think of "select (x by f)", and it bothers me that the meanings are disjoint.
@1084056953 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Aha! The Gwydion documentation is much better: <http://www.gwydiondylan.org/gdref/gdlibs/libs-dylan-dylan.html>
@1084057135 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I find the documentation a little weird, though, in that it says "Gwydion compilers support...", which I guess is true of Mindy, but in the case of d2c it's about what's exported by the Gwydion-supplied dylan module, rather than "the compiler," right?
@1084057154 <housel!housel@cantor.gwydiondylan.org> obviously
@1084057202 <housel!housel@cantor.gwydiondylan.org> "compiler" is being used loosely here
@1084057245 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is subclass() an alias for limited(<class>, subclass-of: <c>) ?
@1084057279 <housel!housel@cantor.gwydiondylan.org> yes
@1084058168 <housel!housel@cantor.gwydiondylan.org> Actually I think subclass() is the primitive nowadays, I doubt limited(<class>) would work
@1084058598 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So that documentation is defunct?
@1084059456 <housel!housel@cantor.gwydiondylan.org> I think maybe it works with Mindy but not d2c
@1084059488 <housel!housel@cantor.gwydiondylan.org> (or dfmc, for that matter)
@1084063163 -kornbluth.freenode.net- hannes (~hannes@pD9E7A611.dip.t-dialin.net) joined the channel
@1084063166 <hannes!~hannes@pD9E7A611.dip.t-dialin.net> re
@1084064444 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Hmm. Melange doesn't handle this correctly:
@1084064467 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   typedef int foo; struct { char foo; }
@1084064498 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It determines that the struct member "foo" is a type, making it a parse error.
@1084065309 <housel!housel@cantor.gwydiondylan.org> are you sure it's not?
@1084065321 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It's a standard system header.
@1084065367 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The data member "foo" is in the namespace of the struct, so x.foo refers to the data member, not the type "foo".
@1084065931 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I mean, the actual code that triggers this is in a standard system header.
@1084065972 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, it looks like fixing it could require percolating some changes up through the grammar. I'll have to stair at it some more.
@1084066044 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/stair/stare
@1084066085 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Or maybe the solution requires introducing namespaces per struct, if it doesn't already do that.
@1084066144 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (In terms of parsing, not Dylan code generation. Melange knows how to generate unique names for the resulting slot accessors.)
@1084067689 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Just did some experimentation and verified that at least one C compiler treats it exactly as a namespace issue. ie., the type "foo" is in the global namespace, but x.foo names "foo" in the struct namespace. The type "foo" does prevent me from using "foo" as a name for something else in the global namespace.
@1084067755 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> typedef int foo; int foo; // not allowed, even though the second "foo" is obviously not a type name
@1084068327 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is there any external documentation for parsergen? Do I need to go read the docs for the Lisp implementation?
@1084071549 -kornbluth.freenode.net- hannes (~hannes@pD9E7A611.dip.t-dialin.net) joined the channel
@1084071551 <hannes!~hannes@pD9E7A611.dip.t-dialin.net> re
@1084077154 -kornbluth.freenode.net- hannes (~hannes@pD9E7A611.dip.t-dialin.net) joined the channel
@1084077155 <hannes!~hannes@pD9E7A611.dip.t-dialin.net> re
@1084079683 <housel!housel@cantor.gwydiondylan.org> cpage: you could read the literate source of Son of Parsergen, found on the Monday website
@1084080455 -kornbluth.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084085767 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> housel: I found an info-dylan message from you about Son of Parsergen, as it happens. Is /src/tools/parsergen/ Son of Parsergen?
@1084088585 -kornbluth.freenode.net- brucehoult (~bruce@202-0-61-59.adsl.paradise.net.nz) joined the channel
@1084093383 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, the C grammars I've found just use IDENTIFIER, whereas Melange uses both IDENTIFIER and TYPEDEF and only allows IDENTIFIER for struct member names.
@1084093467 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm trying to decide whether to adjust the grammar to allow either of IDENTIFIER or TYPEDEF in the struct name position, or just have it only deal with identifiers and check whether it's a typedef name elsewhere in the code.
@1084093740 -kornbluth.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084093745 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084093751 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> hi
@1084093772 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> NickServ says that IDNETIFY is an unrecognised message
@1084093776 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Chris.
@1084093861 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I wonder why it would complain about "IDNETIFY".
@1084098060 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Well I can see why, but I actually typed IDENTIFY . :-)
@1084098071 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> :-)
@1084098082 -kornbluth.freenode.net- activist (~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com) joined the channel
@1084098105 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> is there a mod_dylan for Apache available?
@1084098249 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Andreas is working on one IIRC.
@1084098278 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> status? (beta, dev, cruchy ;-) ...)
@1084098288 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> ^cruchy/crunchy
@1084098323 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION goes to bed to prevent keyboard marks on his forehead
@1084098332 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Erm. I don't know, you'd need to ask him (or someone better informed than I :-) ). There's a Dylan CGI library, and the Koala web server written in Dylan with its Dylan Server Page system.
@1084098337 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> G'night Chris.
@1084098381 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> Contact? (I'm not a Dylanist - yet. Chris has been lobbying, so I'm checking it out)
@1084098391 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Redarding mod_dylan, ask on the gd-hackers mailing list or try later when both the US and Europe are up. Erm, again. :-)
@1084098416 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> I'm looking for a way to use Dylan in an Apache setup (not as an alternate server like Koala).
@1084098444 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> OK. There doesn't seem to be anything in CVS for mod_dylan. You could try out with the CGI lib.
@1084098463 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> Thanks for the pointers - I'll look around later.
@1084098463 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, try to stick to a wider audience if possible. comp.lang.dylan is ideal, then info-dylan, then gd-hackers for Gwydion-specific discussion. I think it's important to maximize visibility.
@1084098471 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The list is on http://www.gwydiondylan.org/cgi-bin/mailman/listinfo/gd-hackers .
@1084098480 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> thx.
@1084098517 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Or Peter (housel) will probably know more know if Andreas isn't around later (they're two of our most core people).
@1084098521 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> OK.
@1084098685 <activist!~activist@dsl-dt-208-38-4-i187-cgy.nucleus.com> thx for the pointers. bye.
@1084100906 -kornbluth.freenode.net- pritlove (~pritlove@73.101.202.62.cust.bluewin.ch) joined the channel
@1084110017 -kornbluth.freenode.net- pritlove (~pritlove@114.123.77.83.cust.bluewin.ch) joined the channel
@1084112072 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084119166 -kornbluth.freenode.net- pritlove (~pritlove@203.117.76.83.cust.bluewin.ch) joined the channel
@1084127887 -kornbluth.freenode.net- hannes (~hannes@pD9E7A611.dip.t-dialin.net) joined the channel
@1084129066 -kornbluth.freenode.net- ^self (~fn@66.209.66.105) joined the channel
@1084129103 <^self!~fn@66.209.66.105> hi
@1084133850 <hannes!~hannes@pD9E7A611.dip.t-dialin.net> hi ^self
@1084145491 -kornbluth.freenode.net- hannes_ (~hannes@pD9E7A902.dip.t-dialin.net) joined the channel
@1084152440 -dircproxy- You disconnected
@1084154462 -dircproxy- You connected
@1084165434 -dircproxy- You disconnected
@1084167550 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084170932 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084171678 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> e
@1084171680 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084172659 -kornbluth.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1084172662 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084172707 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Evening!
@1084172714 <andreas!andreas@cantor.gwydiondylan.org> Hi Chris!
@1084174663 -kornbluth.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084174707 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I fired up my Newton 2000 and I'm reading some notes I wrote about Dylan in 1996 just after I joined Harlequin. It's interesting to see that they're still worth answering.
@1084174723 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/notes/questions
@1084174802 <andreas!andreas@cantor.gwydiondylan.org> chris: put them on a web page. :)
@1084174808 [andreas!andreas@cantor.gwydiondylan.org] ACTION has to leave now, later!
@1084174855 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Does \let or \local have a value?"
@1084175011 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Concrete subclasses of <array> and vector>, and the types of elements of <unicode-string> and <byte-string> are essentially unspecified..."
@1084175211 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Can macros define macros?"
@1084175221 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Some of this stuff might already be in a FAQ.
@1084175272 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Some I should probably post for comment on c.l.d and publish the questions and answers afterwards.
@1084179668 -kornbluth.freenode.net- pritlove (~pritlove@194.95.203.213) joined the channel
@1084180664 <gabor!root@cantor.gwydiondylan.org> re.
@1084180678 <gabor!root@cantor.gwydiondylan.org> chris: publish it in the wiki
@1084180710 <gabor!root@cantor.gwydiondylan.org> Can macros define macros? ---> sure, why not?
@1084180732 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It wasn't clear at the time from my reading of the DRM.
@1084180856 <gabor!root@cantor.gwydiondylan.org> but d2c probably cannot handle this yet
@1084181983 <^self!~fn@66.209.66.105> hi.
@1084197931 -kornbluth.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084197933 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1084198589 -kornbluth.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084198677 -kornbluth.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084200525 -dircproxy- You connected
@1084201052 <^self!~fn@66.209.66.105> my pearpc attempt with macppccd.iso:
@1084201052 <^self!~fn@66.209.66.105> http://web.nilpotent.org/tmp/root1.png
@1084201052 <^self!~fn@66.209.66.105> a few seconds later:
@1084201052 <^self!~fn@66.209.66.105> http://web.nilpotent.org/tmp/output.txt
@1084201559 <gabor!root@cantor.gwydiondylan.org> what is pearpc?
@1084204267 -dircproxy- You disconnected
@1084206136 -kornbluth.freenode.net- pritlove (~pritlove@dsl-082-082-097-189.arcor-ip.net) joined the channel
@1084206955 -dircproxy- You connected
@1084207050 -dircproxy- You disconnected
@1084207093 -dircproxy- You connected
@1084209375 -kornbluth.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084209503 -kornbluth.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) joined the channel
@1084209707 -kornbluth.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1084214343 -kornbluth.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084215951 -dircproxy- You disconnected
@1084222385 -dircproxy- You connected
@1084222556 <housel!housel@cantor.gwydiondylan.org> re
@1084226557 -kornbluth.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084227249 -kornbluth.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084233631 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084235756 <bhoult!~bhoult@morannon.eservglobal.co.nz> http://en.wikipedia.org/wiki/Dylan_programming_language
@1084235764 <bhoult!~bhoult@morannon.eservglobal.co.nz> there are a few things here that should be corrected
@1084235789 <bhoult!~bhoult@morannon.eservglobal.co.nz> no links to Gwydion!
@1084235808 <bhoult!~bhoult@morannon.eservglobal.co.nz> init-keyword does *not* set getter and setter names
@1084235832 <bhoult!~bhoult@morannon.eservglobal.co.nz> most Lisps are *not* interpreters
@1084235854 <bhoult!~bhoult@morannon.eservglobal.co.nz> (but it's hard to generate tight compiled code)
@1084235945 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 'The primary advantage of Lisp-like systems is that everything in the system, including the language itself, can be modified in the language.' -- uh, that depends on having a CLOS-based system with a MOP, which not even the CL standard mandates...
@1084235958 <bhoult!~bhoult@morannon.eservglobal.co.nz> right
@1084236000 <bhoult!~bhoult@morannon.eservglobal.co.nz> w.color-setter(<color>.blue)   means something *totally* different in Dylan!
@1084236095 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> '...whereas in Dylan this completely unrelated concept is completely unrelated.'
@1084236100 [housel!housel@cantor.gwydiondylan.org] ACTION tries to remember his wikipedia login...
@1084236106 <bhoult!~bhoult@morannon.eservglobal.co.nz> not all *that* happy with the general tone in paces either, but factual accuracy should at least be possible
@1084236155 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> <color>.blue?  Huh?
@1084236181 <bhoult!~bhoult@morannon.eservglobal.co.nz> yuh
@1084236213 <housel!housel@cantor.gwydiondylan.org> looks like history doesn't go back far enough to find out who is responsible for these errors
@1084236344 <bhoult!~bhoult@morannon.eservglobal.co.nz> I"m not sure the aspect stuff makes sense either, but then I feel that way about aspect-oriented waffle in general
@1084236971 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> AOP is very similar to exceptions. They allow you to enforce policies without spreading code all over the place to implement them.
@1084237009 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ie., the opposite of AOP and exceptions is having every function return an error result and pass along the results of all functions it calls.
@1084237047 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...or having every function call a "before" and an "after" function around every function they call.
@1084237157 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I agree about the Wikipedia page tone. Wikipedia is generally very neutral when discussing hot topics like "syntax" and "dynamic vs. static". That page seems a little angry.
@1084237188 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Subjective?  Where??
@1084237289 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Even this quote right at the start of the entry seems contentious: " It was originally intended for use with Apple's Newton computer, but their implementation did not reach sufficient maturity in time, and they instead developed NewtonScript for that project."
@1084237332 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "developers avoided it even at the $29 price"  Well, duh! It was unfinished and unsupported.
@1084237368 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That makes it seem like Apple couldn't sell Apple Dylan even at a cheap price, reflecting poorly on Dylan.
@1084237400 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Argh!  That reminds me that I _really_ ought to get around to trying out the copy of Apple Dylan that Felix sent me.
@1084237413 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What copy is that? The TR?
@1084237434 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Yes, I believe so.
@1084237456 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "...and a university implemented..." Do you think it would kill them to mention CMU by name?
@1084237477 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Of course it would.
@1084237488 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Isn't that obvious?
@1084237512 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION chuckles a little
@1084237551 [housel!housel@cantor.gwydiondylan.org] ACTION remembers the trouble he went through to buy a copy of the Apple Dylan TR while living in Taiwan
@1084237562 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think "terribly confusing syntax that most LISP systems use" is overdoing it.
@1084237568 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Indeed.
@1084237586 <housel!housel@cantor.gwydiondylan.org> way over the top, in fact
@1084237871 <housel!housel@cantor.gwydiondylan.org> "were quite good"? last I checked they still are
@1084238135 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Besides, "most"?
@1084238510 -dircproxy- You disconnected
@1084240010 -dircproxy- You connected
@1084252521 -dircproxy- You disconnected
@1084253977 -dircproxy- You connected
@1084262329 -kornbluth.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084266717 -kornbluth.freenode.net- pritlove (~pritlove@194.95.203.200) joined the channel
@1084267462 -kornbluth.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084267462 -kornbluth.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1084272542 -dircproxy- You disconnected
@1084274875 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084275033 <^self!~fn@66.209.66.105> hi
@1084275866 <^self!~fn@66.209.66.105> http://www.omino.com/~poly/art/monominoes/movies/k20040509.mov -- 2mb
@1084275998 [andreas!andreas@cantor.gwydiondylan.org] ACTION launches mplayer.
@1084276446 -dircproxy- You connected
@1084278867 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1084283466 -kornbluth.freenode.net- pritlove_ (~pritlove@194.95.203.200) joined the channel
@1084286456 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084290095 -dircproxy- You disconnected
@1084293185 -dircproxy- You connected
@1084293303 -kornbluth.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084299088 -kornbluth.freenode.net- pritlove (~pritlove@dsl-213-023-156-188.arcor-ip.net) joined the channel
@1084299898 -kornbluth.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084299899 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084300936 -kornbluth.freenode.net- hannes (~hannes@pD9E7B389.dip.t-dialin.net) joined the channel
@1084300938 <hannes!~hannes@pD9E7B389.dip.t-dialin.net> re
@1084301575 -kornbluth.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1084302644 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084302880 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Andreas.
@1084302888 <andreas!andreas@cantor.gwydiondylan.org> Hey Rob!
@1084302904 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Someone was asking after a mod_dylan on Sunday, did they catch you?
@1084303441 <andreas!andreas@cantor.gwydiondylan.org> Nope, not yet.
@1084305555 -kornbluth.freenode.net- pritlove (~pritlove@pD9E7B389.dip.t-dialin.net) joined the channel
@1084306367 -kornbluth.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084307660 -kornbluth.freenode.net- cpage (~Chris@67.118.119.204) joined the channel
@1084307688 <housel!housel@cantor.gwydiondylan.org> hi chris
@1084308111 <andreas!andreas@cantor.gwydiondylan.org> Hello Chris, Peter.
@1084308146 <cpage!~Chris@67.118.119.204> hi
@1084308176 <cpage!~Chris@67.118.119.204> Yeah, someone I know wanted to know if there was an Apache mod_dylan. What would it take to make one?
@1084308208 <housel!housel@cantor.gwydiondylan.org> one could use mod_lisp I think
@1084308210 <andreas!andreas@cantor.gwydiondylan.org> There's a hack to turn a Dylan application into a shared library with a defined entry point.
@1084308213 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> What would such a mod_dylan do?
@1084311332 <cgay!~cgay@ita4fw1.itasoftware.com> the same thing most other apache mods do, stay resident in mem to prevent the overhead of loading on every request?
@1084311347 <andreas!andreas@cantor.gwydiondylan.org> Hi Carl.
@1084311373 <cgay!~cgay@ita4fw1.itasoftware.com> also it could be easier to convince management to go with mod_foo + apache than, say, Koala
@1084311380 <cgay!~cgay@ita4fw1.itasoftware.com> hi andreas
@1084311469 <cgay!~cgay@ita4fw1.itasoftware.com> were you actually working on mod_dylan?
@1084311548 <cgay!~cgay@ita4fw1.itasoftware.com> does NZ have koalas?  or just AU?
@1084311601 <andreas!andreas@cantor.gwydiondylan.org> What I did was implementing support for producing shared libraries. There's no way to create predictable names for C convention call entries yes.
@1084311619 <housel!housel@cantor.gwydiondylan.org> yet
@1084311651 <andreas!andreas@cantor.gwydiondylan.org> It's a SMOP. :)
@1084311664 <hannes!~hannes@pD9E7B389.dip.t-dialin.net> andreas: smop?
@1084311671 <cgay!~cgay@ita4fw1.itasoftware.com> simple matter of programming
@1084311687 <cgay!~cgay@ita4fw1.itasoftware.com> unfortunately we have so many SMOPs :)
@1084311731 <housel!housel@cantor.gwydiondylan.org> cgay: Could you add any of the Fun-O people who might care to the CC: list of bug 6986?  That's the bug I'm using to track the new splash screen, copyright dates changes, and edition-based restrictions changes
@1084311733 <cgay!~cgay@ita4fw1.itasoftware.com> well, _you_, really.  seeing as how i'm not qualified for most of the low-level hacking that needs to be done
@1084311794 <housel!housel@cantor.gwydiondylan.org> I'm currently working on a list of all of the things in the code that differ between basic/enhanced/internal editions
@1084311965 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The technique I used for exporting entry points was to make a C global for the entry point, get it with the Melange interfaces, then set it to a callback entry. Voila, predictable entry point naming (as long as you start the Dylan runtime up).
@1084312162 <housel!housel@cantor.gwydiondylan.org> By the way, some of the weirdness in the handling of warning/error messages is because we have been building internal editions
@1084312249 <housel!housel@cantor.gwydiondylan.org> (the linux betas were internal editions too).
@1084312401 <bhoult!~bhoult@morannon.eservglobal.co.nz> cgay: NZ only has Koalas in the zoos
@1084312415 <andreas!andreas@cantor.gwydiondylan.org> Morning Bruce!
@1084312427 <bhoult!~bhoult@morannon.eservglobal.co.nz> we do have some wallabies in the wild -- introduced by man, and doing well.
@1084312432 <bhoult!~bhoult@morannon.eservglobal.co.nz> hi Andreas!
@1084312453 <CIA-2!micah@hyperreal.info> commit by andreas to src: Bug: 7005
@1084312453 <CIA-2!micah@hyperreal.info> A <magic-c-representation> is a standin for representations we know
@1084312453 <CIA-2!micah@hyperreal.info> nothing but the C name of.
@1084312463 <andreas!andreas@cantor.gwydiondylan.org> Can't locate /var/lib/bugzilla/data/versioncache at
@1084312463 <andreas!andreas@cantor.gwydiondylan.org>         /usr/share/bugzilla/lib/globals.pl line 636 (#1)
@1084312466 <andreas!andreas@cantor.gwydiondylan.org> Eek!
@1084312472 <bhoult!~bhoult@morannon.eservglobal.co.nz> that again :-(
@1084312492 <bhoult!~bhoult@morannon.eservglobal.co.nz> when someone edits stuff in bugzilla, that file is regenerated by root, without world read permissions
@1084312522 <bhoult!~bhoult@morannon.eservglobal.co.nz> I've been wondering what the best solution is
@1084312541 <bhoult!~bhoult@morannon.eservglobal.co.nz> perhaps making the cvs checking scripts setuid?
@1084312570 <housel!housel@cantor.gwydiondylan.org> fixing the umask of the generator would be better
@1084312905 <andreas!andreas@cantor.gwydiondylan.org> This time, I get:
@1084312907 <andreas!andreas@cantor.gwydiondylan.org> [Tue May 11 21:59:45 2004] processmail: Error in tempfile() using /var/lib/bugzilla/data/versioncache.XXXXX: Parent directory (/var/lib/bugzilla/data/) is not writable
@1084313025 <bhoult!~bhoult@morannon.eservglobal.co.nz> what was that from?
@1084313030 <bhoult!~bhoult@morannon.eservglobal.co.nz> CIA?
@1084313032 <andreas!andreas@cantor.gwydiondylan.org> [Tue May 11 21:59:45 2004] processmail:  at /usr/share/bugzilla/lib/globals.pl line 529
@1084313076 <bhoult!~bhoult@morannon.eservglobal.co.nz> hmm
@1084313084 <bhoult!~bhoult@morannon.eservglobal.co.nz> this stuff is actually in bugzilla itself, isn't it
@1084313108 -kornbluth.freenode.net- prom_ (~prom@217.13.206.33) joined the channel
@1084313110 <prom_!~prom@217.13.206.33> re
@1084313113 <andreas!andreas@cantor.gwydiondylan.org> prom.
@1084313115 <bhoult!~bhoult@morannon.eservglobal.co.nz> re Prom
@1084313117 <hannes!~hannes@pD9E7B389.dip.t-dialin.net> hi prom
@1084313250 <prom_!~prom@217.13.206.33> my shell server is back so ill let this client running... just waiting for the user occopying my nick to change his nick, should he be there, so i dont have to kick them...
@1084313267 <prom_!~prom@217.13.206.33> s/them/him/ *gaga*
@1084314185 <CIA-2!micah@hyperreal.info> commit by andreas to src: Bug: 7005
@1084314185 <CIA-2!micah@hyperreal.info> Implement passing C structs by value.
@1084314187 <CIA-2!micah@hyperreal.info> commit by andreas to src: Bug: 7006
@1084314189 <CIA-2!micah@hyperreal.info> Porting latest work on GTK2 over from 2.5 branch to HEAD.
@1084314624 -kornbluth.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084314779 -kornbluth.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084314781 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1084314858 <andreas!andreas@cantor.gwydiondylan.org> re hannes.
@1084315984 <CIA-2!micah@hyperreal.info> commit by andreas to src: Bug: 7006
@1084315984 <CIA-2!micah@hyperreal.info> Fix dependency.
@1084316133 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/cgi-bin/bugzilla/show_bug.cgi?id=6986 now contains the full list of what you can and cannot do in various editions of Functional Developer
@1084316290 <housel!housel@cantor.gwydiondylan.org> I think we need to maintain two editions (internal and ~internal), primarily for the debugging/error trapping related stuff
@1084316300 <housel!housel@cantor.gwydiondylan.org> most of the other restrictions can be removed
@1084316867 <CIA-2!micah@hyperreal.info> commit by andreas to src: Bug: 7005
@1084316867 <CIA-2!micah@hyperreal.info> Special case for magic representations.
@1084317499 -kornbluth.freenode.net- pritlove (~pritlove@pD9E7B389.dip.t-dialin.net) joined the channel
@1084317517 <prom!~prom@217.13.206.33> housel: did you succeed in building on windows while i was gone? ill have a windows workstation in a couple of days...
@1084317543 <housel!housel@cantor.gwydiondylan.org> yes, see http://www.gwydiondylan.org/~housel/fundev-snapshot.html
@1084317562 <housel!housel@cantor.gwydiondylan.org> you need VC++6 to use it at the moment
@1084317625 <housel!housel@cantor.gwydiondylan.org> also it sometimes crashes in the garbage collector for reasons I haven't identified yet
@1084317670 <housel!housel@cantor.gwydiondylan.org> (Actually you can use it with VC++7, but the debugging info is incomplete)
@1084317719 <housel!housel@cantor.gwydiondylan.org> I now have my own copy of VC6 ($150 on ebay) so building snapshots should go a bit more smoothly in the near future
@1084317731 <andreas!andreas@cantor.gwydiondylan.org> The command line compiler doesn't seem to crash, so I suspect thread problems.
@1084317736 <prom!~prom@217.13.206.33> housel: cool.
@1084317752 <prom!~prom@217.13.206.33> andreas: *sigh* not again. i still havent solved the problems i had with libgc in framerd.
@1084317777 <prom!~prom@217.13.206.33> andreas: the debugging.html file didnt help... as did the other documentation.
@1084317866 <prom!~prom@217.13.206.33> housel: ah. cool that you own a copy. so we can make public release builds ;)
@1084317922 <housel!housel@cantor.gwydiondylan.org> Once I have the Jam-based build system going, I hope to make everything buildable with MinGW
@1084318158 <prom!~prom@217.13.206.33> housel: wow. didnt notice jam before. the example on the homepage is nice.
@1084318180 <prom!~prom@217.13.206.33> housel: do you intend to let jam know about dylan dependencies? is this easily possible?
@1084318360 <housel!housel@cantor.gwydiondylan.org> For now Jam is going to be used as a replacement for the build-system and linker-support libraries in FunDev, which are responsible for linking object files, running the resource compiler, running the C compiler, etc. as the final stages of a project build
@1084318370 <housel!housel@cantor.gwydiondylan.org> (also copying and installing files)
@1084318386 <andreas!andreas@cantor.gwydiondylan.org> Ok, I have backported the latest GTK2.0 version plus associated modifications to d2c and melange. 
@1084318425 <housel!housel@cantor.gwydiondylan.org> Right now these libraries are a confusing mix of hard-coded logic and *.script files (which contain a confusing array of substitution variables)
@1084318433 <andreas!andreas@cantor.gwydiondylan.org> And will go home and to sleep now. Night!
@1084318445 <housel!housel@cantor.gwydiondylan.org> n8
@1084318728 <prom!~prom@217.13.206.33> housel: yeah, ive seen that.
@1084318776 -kornbluth.freenode.net- pritlove (~pritlove@217.231.156.162) joined the channel
@1084318934 <prom!~prom@217.13.206.33> why does almost everything having to do with building software end up in a confusing nightmarish shell-make-autofoo-dungeon which has to be fixed almost daily? *g*
@1084319006 <housel!housel@cantor.gwydiondylan.org> :-]
@1084319013 -dircproxy- You disconnected
@1084319099 <prom!~prom@217.13.206.33> i mean... the rest of FD is quite nice... it has its dark corners, but nothing is as buggy and hardcoded-supercomplex as the build system.
@1084319432 <prom!~prom@217.13.206.33> but it is a complex problem... too many specific needs, hidden information, hacks.... its kinda like a mirror of the software world out there. building touches a large number of systems, which results in it having to deal with a lot of problems from different systems.
@1084320709 -dircproxy- You connected
@1084320826 <housel!housel@cantor.gwydiondylan.org> using a real scripting language (Jam) should at least make porting and modification a lot more transparent than it is now
@1084322912 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Bug: 7007
@1084322919 <CIA-2!micah@hyperreal.info> Remove these obsolete binaries and include files.
@1084324198 -dircproxy- You disconnected
@1084325777 -kornbluth.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) joined the channel
@1084325821 -kornbluth.freenode.net- alexpotanin (~alex@st-james.mcs.vuw.ac.nz) left the channel
@1084326369 <prom!~prom@217.13.206.33> *lol* i just found out that VT102 terminals scroll faster in the USA than in Europe.
@1084326381 <prom!~prom@217.13.206.33> at least in smooth scrolling mode.
@1084326398 <prom!~prom@217.13.206.33> in the use its 6 lines per second (60 Hz), in europe 5 (50 Hz).
@1084326466 <prom!~prom@217.13.206.33> s/use/us/
@1084333884 -dircproxy- You connected
@1084333904 -zelazny.freenode.net- You joined the channel
@1084333924 <bhoult!~bhoult@morannon.eservglobal.co.nz> re Peter
@1084333960 <housel!housel@cantor.gwydiondylan.org> hi
@1084355941 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.225) joined the channel
@1084367153 <^self!~fn@66.209.66.105> vonnegut: http://www.inthesetimes.com/site/main/article/cold_turkey/ 
@1084375695 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1084376342 <cgay!~cgay@ita4fw1.itasoftware.com> oh man, it's too bad koala/linux isn't ready for prime time.  we have so many web projects to do here and "oddball" languages are no problem
@1084377092 <housel!housel@cantor.gwydiondylan.org> so close... I've spent some time trying to get MPS to work on linux, but so far it keeps crashing
@1084377157 <cgay!~cgay@ita4fw1.itasoftware.com> :(
@1084377191 <housel!housel@cantor.gwydiondylan.org> I'll keep at it
@1084377209 <cgay!~cgay@ita4fw1.itasoftware.com> i'm just glad someone is working on it finally.  i wish i had gotten started earlier on open sourcing
@1084377240 <housel!housel@cantor.gwydiondylan.org> I just updated http://www.gwydiondylan.org/cgi-bin/bugzilla/show_bug.cgi?id=6986 with more info about copyright dates, license texts, and edition-based restrictions
@1084377245 <cgay!~cgay@ita4fw1.itasoftware.com> i think you've made amazing progress on the whole thing
@1084377289 <housel!housel@cantor.gwydiondylan.org> I'd appreciate it if you could add Fun-O people to the CC: list to encourage them to comment
@1084377516 <cgay!~cgay@ita4fw1.itasoftware.com> i found a License.txt in the fun-o sources
@1084377662 -dircproxy- You disconnected
@1084377680 <cgay!~cgay@ita4fw1.itasoftware.com> I attached it to the bug report
@1084378588 <cgay!~cgay@ita4fw1.itasoftware.com> I sent mail to fun-o folks about the bug.  i don't want to add them to the cc list, as it may cause them to get spam (?)
@1084379664 -dircproxy- You connected
@1084379700 <housel!housel@cantor.gwydiondylan.org> ok thanks
@1084379799 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> are you guys getting "rhashem has changed their status to {away, online}" messages? Or is it just my IRC client?
@1084379854 <housel!housel@cantor.gwydiondylan.org> no, I don't see those (from you anyway, just cpage)
@1084379885 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> good
@1084380449 <cgay!~cgay@ita4fw1.itasoftware.com> heh.  i guess perl's version of (ignore-errors ...) is local *CORE::GLOBAL::die = sub {};
@1084380461 <andreas!andreas@cantor.gwydiondylan.org> Hi folks!
@1084380477 <cgay!~cgay@ita4fw1.itasoftware.com> (from a recommended patch to File::Temp to work around Bugzilla errors)
@1084380480 <cgay!~cgay@ita4fw1.itasoftware.com> hi andreas
@1084380484 <andreas!andreas@cantor.gwydiondylan.org> Hi Carl!
@1084381417 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084389244 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084389900 -dircproxy- You disconnected
@1084390108 -zelazny.freenode.net- pritlove (~pritlove@pD9E79CA2.dip.t-dialin.net) joined the channel
@1084393145 -dircproxy- You connected
@1084393851 -zelazny.freenode.net- hannes (~hannes@pD9E79CA2.dip.t-dialin.net) joined the channel
@1084393856 <hannes!~hannes@pD9E79CA2.dip.t-dialin.net> re
@1084393994 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1084394001 <hannes!~hannes@pD9E79CA2.dip.t-dialin.net> hi peter
@1084397067 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084397069 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084397496 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I can die happy
@1084397506 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I found out that kdevelop has a fullscreen mode
@1084397511 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> 1600x1200 of pure code!
@1084397880 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Boot into it and have a codeOS. :-)
@1084397926 <hannes!~hannes@pD9E79CA2.dip.t-dialin.net> rhashem: btw, does kdevelop support dylan highlighting?
@1084398385 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> hannes: nope :(
@1084398464 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> But you can always use the VIM KPart in KDevelop
@1084398515 <housel!housel@cantor.gwydiondylan.org> bleah... I have enough trouble with code that exceeds 80 columns... in full-screen mode who knows what people would do...
@1084398541 <cgay!~cgay@ita4fw1.itasoftware.com> C-x 3
@1084398574 <cgay!~cgay@ita4fw1.itasoftware.com> that's what i would do :)
@1084398608 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> kdevelop (actually, KatePart) has incredibly good wrapping, though
@1084398631 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> it wraps to the last level of indentation, instead of wrapping to the beginning
@1084398691 <bhoult!~bhoult@morannon.eservglobal.co.nz> I hope that 1600x1200 of pure code can be three columns showing different files!
@1084398700 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION normally has 2 x 80 col emacs full-height frames, plus a stack of three xterms on the RHS
@1084398713 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> bhoult: not an a 15" laptop screen...
@1084399387 <bhoult!~bhoult@morannon.eservglobal.co.nz> why not?  15" LCD is pretty similar to 17" CRT in actual size
@1084399525 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> but three columns?
@1084399573 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> 80 lines is a nice ideal, but right now the file i'm looking at has 120
@1084399580 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> err, characters
@1084399635 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> std::cout << foo::bar::baz::message << std::endl --- C++ uses up horizontal space real quick
@1084399636 <bhoult!~bhoult@morannon.eservglobal.co.nz> I've never seen a good reason to go over 80.
@1084399722 <bhoult!~bhoult@morannon.eservglobal.co.nz> your example is less than 80
@1084399746 <bhoult!~bhoult@morannon.eservglobal.co.nz> if you have more things to print then put each one (plus any whitespace/delimiter) on a new line
@1084399763 <bhoult!~bhoult@morannon.eservglobal.co.nz> use'ing some namespaces wouldn't hurt either :-)
@1084399773 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> ideally yes. 
@1084399789 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> In practice, other people aren't so nice to you
@1084399802 <bhoult!~bhoult@morannon.eservglobal.co.nz> then beat them with a clue stick :-)
@1084399844 <bhoult!~bhoult@morannon.eservglobal.co.nz> I don't think 100 or 120 is too excessive...
@1084399887 <bhoult!~bhoult@morannon.eservglobal.co.nz> ... but the single biggest reason I have for sticking religiously to 80 chars is that it allowed you to do 2 column printouts on A4 paper at a readable size, without wrapping/truncation issues
@1084399891 <cgay!~cgay@ita4fw1.itasoftware.com> i think _strictly_ limiting to 80 chars (as in a previous Java-coding job I had) is bad.  Often there are times (error message strings come to mind) when going a bit wider doesn't hurt anything.
@1084399928 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> C++ does make things worse, though. Namespaces have some annoying design deficiencies, such as not having a sane definition of 'using' in header files
@1084399929 <bhoult!~bhoult@morannon.eservglobal.co.nz> error message strings can be triviall split, as C/C++ lexers concatenate adjacent string literals
@1084399941 <cgay!~cgay@ita4fw1.itasoftware.com> hmm.  i never print out code. 
@1084399954 <cgay!~cgay@ita4fw1.itasoftware.com> but i know some people who like to
@1084399958 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> bhoult: remember to use a \ though :)
@1084399965 <bhoult!~bhoult@morannon.eservglobal.co.nz> no!
@1084399970 <bhoult!~bhoult@morannon.eservglobal.co.nz> no \
@1084399982 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> G++ 3.3 doesn't accept it otherwise
@1084399993 <bhoult!~bhoult@morannon.eservglobal.co.nz> printf("this is a"
@1084399999 <bhoult!~bhoult@morannon.eservglobal.co.nz>        "long literal");
@1084400009 <bhoult!~bhoult@morannon.eservglobal.co.nz> oops .. needed a space, though :-)
@1084400024 <cgay!~cgay@ita4fw1.itasoftware.com> heh.  i've run into that bug many times
@1084400077 <bhoult!~bhoult@morannon.eservglobal.co.nz> that g++ 3.3 won't accept is a run-on multi line literal
@1084400082 <bhoult!~bhoult@morannon.eservglobal.co.nz> that's a quite different thing
@1084400087 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. which breaks indentation
@1084400100 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> how so?
@1084400118 <bhoult!~bhoult@morannon.eservglobal.co.nz> because the run-on part starts right from the 1st column
@1084400160 <housel!housel@cantor.gwydiondylan.org> I don't have any problem with code that goes over 80 occasionally, it's mainly code written by people whose windows are so wide they have no idea where 80 is that annoys me
@1084400176 <housel!housel@cantor.gwydiondylan.org> it's usually worse for assembly than for C
@1084400220 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> whoa. Good point. I have no idea how I missed that
@1084400241 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> probably because I've never had occasion to use it that way...
@1084400250 <cgay!~cgay@ita4fw1.itasoftware.com> some editors have a nice feature of drawing a line down the target column (normally 80)
@1084400278 <cgay!~cgay@ita4fw1.itasoftware.com> (at least, i think it's a nice feature, but i've never actually used it so i don't know how annoying it would be)
@1084400297 <housel!housel@cantor.gwydiondylan.org> hmmm... maybe we can implement that in deuce
@1084400337 <hannes!~hannes@pD9E79CA2.dip.t-dialin.net> cgay: hmm, you know if vim has this feature?
@1084400352 <cgay!~cgay@ita4fw1.itasoftware.com> no, never used vim
@1084400367 <bhoult!~bhoult@morannon.eservglobal.co.nz> I do admit that smart wrapping in the editor and/or print program would help a lot
@1084400382 <cgay!~cgay@ita4fw1.itasoftware.com> SlickEdit is where I saw it
@1084400425 <housel!housel@cantor.gwydiondylan.org> the last time I used Emacs FORTRAN mode (1989 or so) it drew a line down column 72
@1084400433 <bhoult!~bhoult@morannon.eservglobal.co.nz> not so much when writing new code, but if you need to work with someone else's excessively wide code
@1084400479 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> bhoult: For long XML documents, a smart wrapping editor is the only way to go
@1084400577 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> otherwise, if you try to keep the XML-style indentation going, you need to keep reformatting your paragraph when you add something in the middle
@1084400639 <housel!housel@cantor.gwydiondylan.org> M-q works fine for me in PSGML-Emacs
@1084400759 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> all statements regarding editors should usually be prefaced with 'excepting emacs'
@1084400770 <housel!housel@cantor.gwydiondylan.org> :-)
@1084400952 <cpage!~Chris@67.118.119.204> Once again, using plain text files for program editing and presentation imposes unfortunate limits on readability.
@1084400974 [cgay!~cgay@ita4fw1.itasoftware.com] ACTION sets his line prefix to 'excepting emacs'
@1084400976 <cpage!~Chris@67.118.119.204> You want line-wrapping to be dynamic.
@1084400995 <housel!housel@cantor.gwydiondylan.org> I do?
@1084401013 <cpage!~Chris@67.118.119.204> Well, either that you want to suffer through code with lines too wide for you to view in some instances. :-)
@1084401049 <cpage!~Chris@67.118.119.204> But more important: We don't want just line wrapping, we need smart line-breaking that maximizes readibility.
@1084401167 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> katepart does that
@1084401171 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> eg: breaks at commas, spaces
@1084401238 <cgay!~cgay@ita4fw1.itasoftware.com> does it work well in practice?  i mean does it get it right all the time?
@1084401257 <cgay!~cgay@ita4fw1.itasoftware.com> i assume if you explicitly enter a \n it respects it?
@1084401371 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I don't follow
@1084401417 <cgay!~cgay@ita4fw1.itasoftware.com> sorry.  do you have to explicitly ask it to do the wrapping, or it does it automatically?
@1084401448 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> it does it whenever the line runs over either the width of the window, or a pre-specified length
@1084401491 <cgay!~cgay@ita4fw1.itasoftware.com> ok.  if you have explicitly typed a Newline will it ever remove that Newline?
@1084401519 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I've never seen it do that
@1084401681 <cgay!~cgay@ita4fw1.itasoftware.com> i c.  so it probably just wraps from the beginning of line to the end of the line when certain chars (Space, Comma) are typed, and doesn't touch any other lines.  I think Chris had in mind something a bit more radical
@1084401950 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I guess
@1084401972 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> Well, bye all. Tomorrow's session: CVS log messages and why you should use them...
@1084401991 <cgay!~cgay@ita4fw1.itasoftware.com> c u
@1084402233 <housel!housel@cantor.gwydiondylan.org> bhoult: can cvszilla automatically resolve bugs based on the log message?
@1084402247 <housel!housel@cantor.gwydiondylan.org> If not would that be a desirable feature?
@1084402948 <bhoult!~bhoult@morannon.eservglobal.co.nz> can you expand on that?
@1084402968 <bhoult!~bhoult@morannon.eservglobal.co.nz> do you mean close the bug if you say "fixed", or the like?
@1084402984 <housel!housel@cantor.gwydiondylan.org> yes
@1084402995 <bhoult!~bhoult@morannon.eservglobal.co.nz> it doesn't do that
@1084403012 <housel!housel@cantor.gwydiondylan.org> "Bug: 7003 (Resolved)" maybe
@1084403038 <bhoult!~bhoult@morannon.eservglobal.co.nz> I expect it could be enhanced to do that easily enough
@1084403063 <bhoult!~bhoult@morannon.eservglobal.co.nz> just a matter of putting the appropriate entries into the appropriate sql tables
@1084403233 <housel!housel@cantor.gwydiondylan.org> Do you think that would be a good idea?
@1084403316 <bhoult!~bhoult@morannon.eservglobal.co.nz> I find that I often want to put a comment into bugzilla when closing a bug anyway -- one separate from the cvs commit comment, espeiclaly if there was more than one cvs commit against the bug
@1084404368 <bhoult!~bhoult@morannon.eservglobal.co.nz> heh.  just asked a few people around here.  Universal reaction was "why would yu want to do that?"
@1084404421 <bhoult!~bhoult@morannon.eservglobal.co.nz> Part of that may be because we give customers access to bugzillla, but have modified it so that cvs commit comments aren't visible to customers, and we have internal bugzilla comments the customer can't see as well.
@1084404445 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. so if the cvs commit closed the bug the person who raised it would only se that it had been closed, not why
@1084404461 <bhoult!~bhoult@morannon.eservglobal.co.nz> none of which applies to gwydion
@1084404524 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084404538 -zelazny.freenode.net- hannes_ (~hannes@pD9E7A98E.dip.t-dialin.net) joined the channel
@1084404544 <hannes_!~hannes@pD9E7A98E.dip.t-dialin.net> re
@1084404611 [hannes_!~hannes@pD9E7A98E.dip.t-dialin.net] ACTION thinks closing bugs with cvs commit is a good idea.
@1084404615 <bhoult!~bhoult@morannon.eservglobal.co.nz> more sensible people might wellof course  make a distinction between a call tracking system and a bug tracking system
@1084404646 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION doesn't object if someone wants to implement it :-)
@1084405104 <bhoult!~bhoult@morannon.eservglobal.co.nz> one problem is you'd want to change both the "resolution" and "bug_status" columns in the bugs table
@1084405105 -zelazny.freenode.net- pritlove_ (~pritlove@pD9E7A98E.dip.t-dialin.net) joined the channel
@1084405120 <housel!housel@cantor.gwydiondylan.org> right
@1084405211 <bhoult!~bhoult@morannon.eservglobal.co.nz> so what you'd probably want to put is one of 'FIXED','INVALID','WONTFIX','DUPLICATE','WORKSFORME' and then set bug_status to 'RESOLVED'
@1084405290 <bhoult!~bhoult@morannon.eservglobal.co.nz> do we have any process for moving from RESOLVED to VERIFIED and/or CLOSED?
@1084405380 <housel!housel@cantor.gwydiondylan.org> not really... at work the release manager does it (back when we used to have one)
@1084405469 <housel!housel@cantor.gwydiondylan.org> I think we have quite a few bugs in RESOLVED/FIXED state that could be CLOSED because we have subsequently done a release
@1084405497 <bhoult!~bhoult@morannon.eservglobal.co.nz> exactly.  here the tester sets VERIFIED and then the customer sets CLOSED when it's installed in production and they're happy with the fix
@1084406535 -zelazny.freenode.net- pritlove (~pritlove@pD9E7A98E.dip.t-dialin.net) joined the channel
@1084408529 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084410052 -zelazny.freenode.net- cpage_ (~Chris@208.240.243.170) joined the channel
@1084412780 -zelazny.freenode.net- cpage__ (~Chris@67.118.119.204) joined the channel
@1084413618 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084414260 -dircproxy- You disconnected
@1084417629 -dircproxy- You connected
@1084430207 -zelazny.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) joined the channel
@1084438644 -zelazny.freenode.net- cpage_ (~Chris@208.240.243.170) joined the channel
@1084438678 <andreas!andreas@cantor.gwydiondylan.org> Moin!
@1084440924 -zelazny.freenode.net- cpage__ (~Chris@67.118.119.204) joined the channel
@1084440963 -zelazny.freenode.net- cpage (~Chris@67.118.119.204) joined the channel
@1084443434 <CIA-2!micah@hyperreal.info> commit by cgay to fundev: job: 7001
@1084443435 <CIA-2!micah@hyperreal.info> "Build" -> "build", where appropriate
@1084444864 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084445254 -zelazny.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1084445670 -zelazny.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084448348 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084448421 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.203) joined the channel
@1084451918 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.203) joined the channel
@1084454117 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.203) joined the channel
@1084454657 <prom!~prom@217.13.206.33> hey channel
@1084454680 <housel!housel@cantor.gwydiondylan.org> hi prom
@1084454911 <andreas!andreas@cantor.gwydiondylan.org> Hi prom, Peter.
@1084456769 <prom!~prom@217.13.206.33>  
@1084456774 <prom!~prom@217.13.206.33> *grrrr*
@1084458740 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1084459785 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1084461449 <cgay!~cgay@ita4fw1.itasoftware.com> does anyone remember if FD has weak tables?  what about GD?
@1084461719 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> http://www.functionalobjects.com/products/doc/core/core_11.htm
@1084461725 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> FunDev seems to have them
@1084461783 <housel!housel@cantor.gwydiondylan.org> GD doesn't
@1084461824 <cgay!~cgay@ita4fw1.itasoftware.com> thanks
@1084463554 -zelazny.freenode.net- wm3 (~wm3@ZF129022.ppp.dion.ne.jp) joined the channel
@1084464217 -dircproxy- You disconnected
@1084464465 <wm3!~wm3@ZF129022.ppp.dion.ne.jp> Hi.
@1084464713 <wm3!~wm3@ZF129022.ppp.dion.ne.jp> Bootstrapping failed.
@1084464838 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084465165 <wm3!~wm3@ZF129022.ppp.dion.ne.jp> . Bye.
@1084467162 -dircproxy- You connected
@1084467235 <housel!housel@cantor.gwydiondylan.org> That was helpful...
@1084468461 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084473114 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1084473360 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) left the channel
@1084474744 -zelazny.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1084474864 -zelazny.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084475533 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084476745 -zelazny.freenode.net- hannes (~hannes@pD9E7A98E.dip.t-dialin.net) joined the channel
@1084476749 <hannes!~hannes@pD9E7A98E.dip.t-dialin.net> re
@1084476974 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1084477035 <housel!housel@cantor.gwydiondylan.org> it looks like MPS on linux is actually working (to some degree) on "real" linux, but isn't working in the FreeBSD linuxulator because the page fault sigcontext isn't being filled in properly
@1084477082 <housel!housel@cantor.gwydiondylan.org> haven't tried multi-threading yet
@1084478795 <housel!housel@cantor.gwydiondylan.org> fortunately fixing the linuxulator looks like a 1-line change
@1084482166 <housel!housel@cantor.gwydiondylan.org> maybe not so easy... the FreeBSD page fault handler throws away some information
@1084482425 <housel!housel@cantor.gwydiondylan.org> I can hack MPS for now though
@1084484051 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084488000 -zelazny.freenode.net- hannes_ (~hannes@pD9E7B9E2.dip.t-dialin.net) joined the channel
@1084488548 -zelazny.freenode.net- hannes__ (~hannes@pD9E7B759.dip.t-dialin.net) joined the channel
@1084489274 <housel!housel@cantor.gwydiondylan.org> aha... an MPS assertion failure while running the compiler
@1084489292 <housel!housel@cantor.gwydiondylan.org> hope it's related to the crashes on windows... should be easier to debug here
@1084489416 <cgay!~cgay@ita4fw1.itasoftware.com> good good...
@1084491100 -zelazny.freenode.net- pritlove (~pritlove@212.222.128.123) joined the channel
@1084493307 -zelazny.freenode.net- hannes (~hannes@pD9E7B759.dip.t-dialin.net) joined the channel
@1084494844 -zelazny.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1084495295 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084495298 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1084495780 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1084495805 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi peter
@1084496929 -zelazny.freenode.net- cpage__ (~Chris@67.118.119.204) joined the channel
@1084497006 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084498169 -dircproxy- You disconnected
@1084501064 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084501102 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> cpage, if you see this, the following line causes dylan-mode to break:
@1084501123 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com>       if (looking-at?("http://", buffer, bpos, qpos | epos))
@1084501148 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> if I remove the :// the problem goes away
@1084501387 <prom!~prom@217.13.206.33> re chan
@1084501393 [prom!~prom@217.13.206.33] ACTION has windows now ;)
@1084501406 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> lucky dog
@1084501437 <prom!~prom@217.13.206.33> too bad i dont have my vs 6.0 yet.
@1084501504 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> did you get a new machine with windows?
@1084501517 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> or installed it on an old box?
@1084501593 <prom!~prom@217.13.206.33> i installed it on my workstation. on a spare disk i didnt need anywhere else.
@1084502609 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> carlgay_: Yeah, I think it may be a general problem with how it parses strings. It's not ignoring stuff inside strings that it should.
@1084502632 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (or maybe I'm thinking of how it parses stuff inside comments incorrectly.)
@1084502662 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Please log a bug about this specific case.
@1084502684 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm doing some work on Melange. Once I'm done there I should have time to look at dylan-mode some more.
@1084503129 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> ok
@1084503462 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION just ordered a new machine
@1084503527 <bhoult!~bhoult@morannon.eservglobal.co.nz> Athlon XP3200+, Asus A7N8X-X MB.  Should be a bit quicker than my current 700 MHz one
@1084505363 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> bugzilla is glacial
@1084505401 <bhoult!~bhoult@morannon.eservglobal.co.nz> more likely to be the net connection
@1084505413 <bhoult!~bhoult@morannon.eservglobal.co.nz> it might be Perl, but it's on a 1.2 GHz machine
@1084505496 <bhoult!~bhoult@morannon.eservglobal.co.nz> hmm
@1084505526 <bhoult!~bhoult@morannon.eservglobal.co.nz> load average is over 5
@1084505551 <bhoult!~bhoult@morannon.eservglobal.co.nz> what the hell is "logtail" and why is it using over a GB of RAM??
@1084505599 <bhoult!~bhoult@morannon.eservglobal.co.nz> well I killed it
@1084505606 <bhoult!~bhoult@morannon.eservglobal.co.nz> we had essentially zero swap space left
@1084505610 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i just found out recently that our bugzilla at work is running on a 400MHz Celeron. :)  it's also glacial
@1084505631 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> thanks
@1084505658 <bhoult!~bhoult@morannon.eservglobal.co.nz> now lpd is using 30% cpu
@1084505661 <bhoult!~bhoult@morannon.eservglobal.co.nz> wtf?
@1084505777 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> good luck and g'night
@1084506800 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Maybe the server needs a cron job that looks for unexpected processes and kills them.
@1084507288 <bhoult!~bhoult@morannon.eservglobal.co.nz> well, someone or something must have started it!
@1084507306 <bhoult!~bhoult@morannon.eservglobal.co.nz> oh f*ck
@1084507312 <bhoult!~bhoult@morannon.eservglobal.co.nz> load avg back up near 7
@1084507315 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And delete their account. :-)
@1084507318 <bhoult!~bhoult@morannon.eservglobal.co.nz> kswapd working hard
@1084507338 <bhoult!~bhoult@morannon.eservglobal.co.nz> logtail back again and up to 239  MB
@1084508059 <bhoult!~bhoult@morannon.eservglobal.co.nz>  /var/log/messages and /var/log/syslog were 1.7 GB!!!  (I've nuked them to reclaim diskspace)
@1084508257 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION thinks tinderbox went nuts somehow, but doesn't understand it
@1084508283 <bhoult!~bhoult@morannon.eservglobal.co.nz> load average now back under 1.0, plenty of free swap and disk
@1084508333 <bhoult!~bhoult@morannon.eservglobal.co.nz> logtail appears to be part of tinderbox.  I suspect it reads the whole syslog into RAM
@1084511351 -dircproxy- You connected
@1084511562 <housel!housel@cantor.gwydiondylan.org> why is the syslog that long?
@1084511566 <housel!housel@cantor.gwydiondylan.org> what was in it?
@1084511575 <bhoult!~bhoult@morannon.eservglobal.co.nz> well I didn't read *all* of it
@1084511594 <bhoult!~bhoult@morannon.eservglobal.co.nz> but if it's still happening we'll soon have some more :-)
@1084511625 <housel!housel@cantor.gwydiondylan.org> heh
@1084511863 <bhoult!~bhoult@morannon.eservglobal.co.nz> meanwhile the system is looking much healthier :-)
@1084514059 <housel!housel@cantor.gwydiondylan.org> debugging garbage collectors is such a pain
@1084521877 <^self!~fn@66.209.66.105> hi.
@1084526726 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084529552 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084529586 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084530950 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084530953 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning
@1084531986 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: around?
@1084532588 -zelazny.freenode.net- pritlove (~pritlove@217.225.3.135) joined the channel
@1084533453 <^self!~fn@66.209.66.105> hmm.
@1084533620 <andreas!andreas@cantor.gwydiondylan.org> +1
@1084536156 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084539832 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084540453 -zelazny.freenode.net- hannes (~hannes@pD9E7B759.dip.t-dialin.net) joined the channel
@1084542159 <cgay!~cgay@ita4fw1.itasoftware.com> hannes: what's up?
@1084542250 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> cgay: about virtualhosting, i already added request-host to request class, so we only have to prefix the document root with request-host.
@1084542328 <cgay!~cgay@ita4fw1.itasoftware.com> I think the main issue is making it all configurable
@1084542454 <cgay!~cgay@ita4fw1.itasoftware.com> I'm planning to make invoke-handler dispatch on <virtual-host> and <request> i think
@1084542670 <cgay!~cgay@ita4fw1.itasoftware.com> actually, your log message showing the Request-Host was what got me started working on it 
@1084542680 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> i like how virtual hosting is done in thttpd, you've one directory for each virtual host which name is the hostname. you can only configure to enable virtual hosting or not, if you enable virtual hosting, it looks in document-root/hostname/
@1084542759 <cgay!~cgay@ita4fw1.itasoftware.com> do they allow for different security policies for different vhosts?
@1084542863 <cgay!~cgay@ita4fw1.itasoftware.com> hey, i found an amusing bug in the directory listing code.  you (or tim?) aren't in a negative timezone offset, i guess :)
@1084542953 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> cgay: yes, we're both in gmt+1/2 (depending on daylight saving). what do you mean with security policies?
@1084543005 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> what we need are configurable status codes (for redirects,...)
@1084543044 <cgay!~cgay@ita4fw1.itasoftware.com> just that apache allows you to specify different authorization types for each vhost.  different password files etc
@1084543150 <cgay!~cgay@ita4fw1.itasoftware.com> i guess thttpd must canonicalize the host name to fqdn before looking up docs
@1084543263 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> apache has for each directory a htacess/passwd. i don't like the idea of htaccess/passwd, one (or more) config file(s) where you specify user + password + url are better (these files should be in the config-directory)
@1084543370 <cgay!~cgay@ita4fw1.itasoftware.com> can you expand on that?  what's the url for?
@1084543579 <cgay!~cgay@ita4fw1.itasoftware.com> isn't that basically what htaccess/passwd _is_? one or more config files where you specific username/password
@1084543589 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> sth like <url="hostname/dir"><user><name>foo</name><pass>bar</pass></user></url>. you need a unique identifier if you have all access lists in one file and not one file per directory
@1084543650 <cgay!~cgay@ita4fw1.itasoftware.com> specific -> specify
@1084543693 <cgay!~cgay@ita4fw1.itasoftware.com> XML is soooo ugly.  i'm of two minds about using it sometimes.
@1084543941 <cgay!~cgay@ita4fw1.itasoftware.com> i don't see anything in the thttpd docs about what happens if the site is reached by an alias hostname
@1084543955 <cgay!~cgay@ita4fw1.itasoftware.com> i guess you just create links to the main dir
@1084543982 <cgay!~cgay@ita4fw1.itasoftware.com> that sounds reasonable
@1084544143 <cgay!~cgay@ita4fw1.itasoftware.com> (will that work on Windows?)
@1084544534 <housel!housel@cantor.gwydiondylan.org> not really
@1084544737 <cgay!~cgay@ita4fw1.itasoftware.com> i really want to keep it cross-platform
@1084546539 <housel!housel@cantor.gwydiondylan.org> my take on xml is that it's The Right Thing for documents, but overkill for config files or communications protocols
@1084546869 <cgay!~cgay@ita4fw1.itasoftware.com> sounds reasonable
@1084547122 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084548601 <CIA-1!micah@hyperreal.info> commit by cgay to libraries: job:minor
@1084548601 <CIA-1!micah@hyperreal.info> minor mods
@1084548688 <CIA-1!micah@hyperreal.info> commit by cgay to libraries: job:minor
@1084548688 <CIA-1!micah@hyperreal.info> Fix minor bugs in directory listing
@1084548793 <CIA-1!micah@hyperreal.info> commit by cgay to libraries: job: 7010
@1084548793 <CIA-1!micah@hyperreal.info> checkpoint vhost.dylan (not yet included in build)
@1084548818 <cgay!~cgay@ita4fw1.itasoftware.com> versioncache schmersioncache
@1084548847 -dircproxy- You disconnected
@1084549005 <andreas!andreas@cantor.gwydiondylan.org> Hi Carl!
@1084549142 <andreas!andreas@cantor.gwydiondylan.org> I've done some more research on the Linux FD crashes. It happens when a new thread is created while another one collects garbage. I'm not entirely sure what the mechanics are yet, but it might have to do with recent design changes in glibc, in the form of a "manager" thread that's actually responsible for spawning new threads.
@1084549194 <andreas!andreas@cantor.gwydiondylan.org> Both the GC'ing thread and the manager thread throw around signals. There's core running where Boehm expects none to be active.
@1084549200 <andreas!andreas@cantor.gwydiondylan.org> s/core/code/.
@1084549216 <andreas!andreas@cantor.gwydiondylan.org> I'll try to add some gratitious locking to Boehm. Maybe this helps.
@1084549271 [andreas!andreas@cantor.gwydiondylan.org] ACTION is going to CCC, brb.
@1084549308 <cgay!~cgay@ita4fw1.itasoftware.com> hi andreas
@1084551287 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084551449 -dircproxy- You connected
@1084551501 <housel!housel@cantor.gwydiondylan.org> hi andreas
@1084551553 <andreas!andreas@cantor.gwydiondylan.org> Hi Peter.
@1084551593 <andreas!andreas@cantor.gwydiondylan.org> Garbage collector problems are tough, I agree. What's your state with MPS on Linux?
@1084551646 <housel!housel@cantor.gwydiondylan.org> I've made quite a bit of progress, but it's dieing with an assertion failure in the middle of a collection
@1084551670 <housel!housel@cantor.gwydiondylan.org> while running the compiler
@1084551711 <andreas!andreas@cantor.gwydiondylan.org> Related to the problems we see on Win32, or something different?
@1084551715 <housel!housel@cantor.gwydiondylan.org> one part of MPS thinks something should be an object, but it no longer is
@1084551738 <housel!housel@cantor.gwydiondylan.org> I think it's not unlikely that they are the same problem
@1084551774 <housel!housel@cantor.gwydiondylan.org> I hope so, because my un*x low-level debugging skills are much better than my corresponding win32 skills
@1084551883 <andreas!andreas@cantor.gwydiondylan.org> Hm, the tools available for win32 are certainly easier to use than gdb.
@1084551935 <housel!housel@cantor.gwydiondylan.org> MSVC doesn't like FunDev symbol tables though
@1084552039 <housel!housel@cantor.gwydiondylan.org> I could debug FunDev with itself, but the debugger doesn't get along with the "cold" MPS due to the stack check
@1084552088 <andreas!andreas@cantor.gwydiondylan.org> I think I could get the authors of OllyDbg (http://home.t-online.de/home/Ollydbg/) to include support for weird symbol tables.
@1084552177 <housel!housel@cantor.gwydiondylan.org> nifty
@1084552850 [andreas!andreas@cantor.gwydiondylan.org] ACTION is reading MPS documentation.
@1084552994 [housel!housel@cantor.gwydiondylan.org] ACTION is writing a message asking people to fix the linuxulator
@1084553840 <andreas!andreas@cantor.gwydiondylan.org> I'm deeply impressed by MPS. Lots of people must have thought really hard for a long time about this.
@1084553953 <housel!housel@cantor.gwydiondylan.org> yes... it is quite impressive... I just wish I could admire it from a distance and not have to try to debug its innards :-)
@1084554200 <andreas!andreas@cantor.gwydiondylan.org> Oh, you will learn more about it this way. :)
@1084556156 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084556851 <housel!housel@cantor.gwydiondylan.org> gah... I dropped run-time/pentium-linux from -O2 to -O1 and it worked
@1084556882 <andreas!andreas@cantor.gwydiondylan.org> *sigh*
@1084556896 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> gcc is great.
@1084556901 <hannes!~hannes@pD9E7B759.dip.t-dialin.net> :(
@1084556907 <andreas!andreas@cantor.gwydiondylan.org> What version of MPS were you using?
@1084556922 <housel!housel@cantor.gwydiondylan.org> cold (ci)
@1084556944 <andreas!andreas@cantor.gwydiondylan.org> 1.100.1, or the one you downloaded with wget?
@1084556948 <housel!housel@cantor.gwydiondylan.org> the latter
@1084556986 <andreas!andreas@cantor.gwydiondylan.org> Ok, I'll try 1.100.1, and your version if it fails.
@1084557016 <housel!housel@cantor.gwydiondylan.org> you want a diff?
@1084557026 <andreas!andreas@cantor.gwydiondylan.org> Yep.
@1084557039 <housel!housel@cantor.gwydiondylan.org> (to the run-time, I mean)
@1084557091 <andreas!andreas@cantor.gwydiondylan.org> You could as well commit it, if it doesn't break the Boehm variant.
@1084557104 <housel!housel@cantor.gwydiondylan.org> if it didn't, I would
@1084557145 <housel!housel@cantor.gwydiondylan.org> (need to make it configurable, or (ideally) build two libdylan.so's, one for each gc)
@1084557199 <andreas!andreas@cantor.gwydiondylan.org> If MPS works, would we want to continue supporting Boehm?
@1084557278 <housel!housel@cantor.gwydiondylan.org> yes, so people can develop non-free software
@1084557290 <andreas!andreas@cantor.gwydiondylan.org> Ok.
@1084557298 <housel!housel@cantor.gwydiondylan.org> (not that I'd encourage them to do that, but...)
@1084557304 <andreas!andreas@cantor.gwydiondylan.org> Do you happen to know how to regenerate runtime.s?
@1084557317 <housel!housel@cantor.gwydiondylan.org> yes
@1084557329 <housel!housel@cantor.gwydiondylan.org> pentium-linux-rtg
@1084557365 <housel!housel@cantor.gwydiondylan.org> later I'll fix the bootstrap process to build that and use it
@1084557391 <andreas!andreas@cantor.gwydiondylan.org> I have my doubts about the quality of the get_current_teb implementation.
@1084557436 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/~housel/run-time.diff
@1084557988 <andreas!andreas@cantor.gwydiondylan.org> Why did you disable the sigactions?
@1084559095 -zelazny.freenode.net- maestro (~vitaminn@clusterfw.beeline3G.ru) joined the channel
@1084559219 -zelazny.freenode.net- maestro (~vitaminn@clusterfw.beeline3G.ru) left the channel
@1084559490 <housel!housel@cantor.gwydiondylan.org> I don't remember now why I did that... it wasn't working in the linuxulator, maybe
@1084559577 <andreas!andreas@cantor.gwydiondylan.org> Ok. It seems to work for me so far. There is a weird problem with the sleep() function: it sleeps way too long. Probably a miscalculation in case of interrupted sleeps (which happens all of the time due to signal handlers for the garbage collector).
@1084559690 <housel!housel@cantor.gwydiondylan.org> you built koala?
@1084559718 <andreas!andreas@cantor.gwydiondylan.org> No, gctest. koala is a good idea, wait...
@1084559772 <andreas!andreas@cantor.gwydiondylan.org> xfd: fmtdy.c:121: dylan_wrapper_check: Assertion `ww[0] == w[0]' failed.
@1084559772 <andreas!andreas@cantor.gwydiondylan.org> Aborted
@1084559801 <housel!housel@cantor.gwydiondylan.org> bummer... that's the same problem I've been seeing
@1084559848 <housel!housel@cantor.gwydiondylan.org> it's trying to scan or skip an object that has been overwritten
@1084559891 <andreas!andreas@cantor.gwydiondylan.org> Which went away after reducing optimizer settings?
@1084559904 <housel!housel@cantor.gwydiondylan.org> for my limited test, yes
@1084559937 <housel!housel@cantor.gwydiondylan.org> I'm going to try a full compiler bootstrap, which should exercise it a lot more
@1084560137 <andreas!andreas@cantor.gwydiondylan.org> I understand the problem with sleep(). It is implemented using nanosleep, which returns a remaining time in case it gets interrupted by a signal, returning EINTR.
@1084560182 <andreas!andreas@cantor.gwydiondylan.org> The returned value is rounded up to the next multiple of 1/HZ, which is 10 ms.
@1084560213 <andreas!andreas@cantor.gwydiondylan.org> Now if you receive lots of signals, as is the case with the barriers in the garbage collector, the error adds up.
@1084560224 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084560230 <andreas!andreas@cantor.gwydiondylan.org> Up to 260 seconds for 1 second sleep in my test case.
@1084560245 <housel!housel@cantor.gwydiondylan.org> so we should include our own sleep() implementation
@1084560288 <andreas!andreas@cantor.gwydiondylan.org> You'd have to pull the remaining wait time from some other time source.
@1084560335 <andreas!andreas@cantor.gwydiondylan.org> Implementation is in linux-threads-primitives.c, function primitive_sleep().
@1084560337 <housel!housel@cantor.gwydiondylan.org> you could store an absolute start time in a thread-local variable, and base the nanosleep request time on the difference
@1084560839 <andreas!andreas@cantor.gwydiondylan.org> I've filed a bug report.
@1084561197 <housel!housel@cantor.gwydiondylan.org> I'm seeing the failure too with a heavier load (minimal-console-compiler compiling dylan)
@1084561212 <housel!housel@cantor.gwydiondylan.org> I'll keep at it
@1084561634 <andreas!andreas@cantor.gwydiondylan.org> Ok, so I now have two hard GC problems to choose from...
@1084561694 <housel!housel@cantor.gwydiondylan.org> :-)
@1084562052 <andreas!andreas@cantor.gwydiondylan.org> Oh, and gctest becomes incredibly slow after awhile.
@1084562164 <housel!housel@cantor.gwydiondylan.org> one of the changes in my test (the additional calls to dylan_wrapper_check() at allocation time) makes things slower than necessary... I recommend removing those, since it's obviously not an allocation-time problem
@1084562178 <housel!housel@cantor.gwydiondylan.org> s/in my test/in my patch/
@1084562191 -zelazny.freenode.net- _8jean (~user@wallstrasse.knowledgetools.de) joined the channel
@1084562225 <_8jean!~user@wallstrasse.knowledgetools.de> Wow, has this channel grown bigger?
@1084562246 <andreas!andreas@cantor.gwydiondylan.org> Hi 8jean. Yes, it seems. :)
@1084562310 <_8jean!~user@wallstrasse.knowledgetools.de> Good to see.  Many check-in messages, too.
@1084562377 <_8jean!~user@wallstrasse.knowledgetools.de> (Well, perhaps not as many as would be optimal, but more active again.)
@1084562880 <housel!housel@cantor.gwydiondylan.org> gabor has been idle of late... more activity from him would be nice
@1084562915 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> e
@1084562918 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084563056 <housel!housel@cantor.gwydiondylan.org> hi chris
@1084563078 <andreas!andreas@cantor.gwydiondylan.org> Hi Chris!
@1084563290 <_8jean!~user@wallstrasse.knowledgetools.de> Is anyone actually _using_ Dylan here?
@1084563312 <_8jean!~user@wallstrasse.knowledgetools.de> Other than "I wrote a 100-liner three years ago for a company that didn't know what I was doing". :-)
@1084563359 <andreas!andreas@cantor.gwydiondylan.org> I know that there's a substantial project being run in Israel, using Fundev.
@1084563362 <_8jean!~user@wallstrasse.knowledgetools.de> Anyone running a webserver?  Rewriting CIA in Dylan?
@1084563428 <housel!housel@cantor.gwydiondylan.org> as soon as we can get the gc fixed we can do just that
@1084563479 <cgay!~cgay@ita4fw1.itasoftware.com> what does CIA stand for again?
@1084563524 <housel!housel@cantor.gwydiondylan.org> it's a pun on the Central Intelligence Agency
@1084563533 <housel!housel@cantor.gwydiondylan.org> "A mindless robot that watches subversion"
@1084563552 <housel!housel@cantor.gwydiondylan.org> (the first version was done for the picogui svn repository)
@1084563557 <_8jean!~user@wallstrasse.knowledgetools.de> Mr. Tenet looking for WMD on the CVS server?
@1084563560 <cgay!~cgay@ita4fw1.itasoftware.com> i c :)
@1084563608 <andreas!andreas@cantor.gwydiondylan.org> Ok, my SO is waiting for me. See you!
@1084563614 <cgay!~cgay@ita4fw1.itasoftware.com> i thought it also meant "CVS I-something A-something"
@1084563669 <housel!housel@cantor.gwydiondylan.org> a retcon I'm sure
@1084564271 <housel!housel@cantor.gwydiondylan.org> hmmm... for me the failure always seems to happen in primitive-make-closure, and some of the offending objects seem to be closures... I wonder if that's a coincidence
@1084565236 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084565237 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084565262 <housel!housel@cantor.gwydiondylan.org> hi rob
@1084565273 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Peter.
@1084565335 -zelazny.freenode.net- _8jean (~user@wallstrasse.knowledgetools.de) left the channel
@1084567803 <housel!housel@cantor.gwydiondylan.org> hmmm... primitive_copy_r() initializes the header portion of the closure, and sets the size field, but leaves the repeated slots uninitialized
@1084567849 <housel!housel@cantor.gwydiondylan.org> ... and then commits the whole thing
@1084567855 <housel!housel@cantor.gwydiondylan.org> looks suspicious to me
@1084567893 <cgay!~cgay@ita4fw1.itasoftware.com> peter, have you been in contact with Ravenbrook to see if they have any ideas?
@1084568009 <housel!housel@cantor.gwydiondylan.org> not about this
@1084568057 <housel!housel@cantor.gwydiondylan.org> it took them two weeks to get to my last e-mail, I want to get thoroughly and undeniably stuck before I bother them
@1084568084 <cgay!~cgay@ita4fw1.itasoftware.com> well, that's a laudible goal.  ;-)
@1084568169 -dircproxy- You disconnected
@1084569323 -dircproxy- You connected
@1084572055 <housel!housel@cantor.gwydiondylan.org> I added code to fill the closure variable slots with NULL before it commits... that's *way* less efficient than only initializing it once, but it actually seems to work
@1084572093 <housel!housel@cantor.gwydiondylan.org> happy happy happy
@1084572143 <housel!housel@cantor.gwydiondylan.org> I can make it efficient later
@1084572169 -dircproxy- You disconnected
@1084572243 <cgay!~cgay@ita4fw1.itasoftware.com> cool!
@1084572305 -dircproxy- You connected
@1084572315 -dircproxy- You disconnected
@1084574306 -dircproxy- You connected
@1084574647 -zelazny.freenode.net- hannes__ (~hannes@pD9E7B555.dip.t-dialin.net) joined the channel
@1084581061 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084582830 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084585828 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/~housel/gcbench has some benchmark results
@1084585849 <housel!housel@cantor.gwydiondylan.org> it's a little slower due to the overhead of the page fault handler, but not much
@1084585898 <housel!housel@cantor.gwydiondylan.org> (here it = MPS)
@1084587591 -zelazny.freenode.net- iehon (~char@vsat-148-63-65-106.c002.t7.mrt.starband.net) joined the channel
@1084587765 <housel!housel@cantor.gwydiondylan.org> hi
@1084587796 <iehon!~char@vsat-148-63-65-106.c002.t7.mrt.starband.net> Hi.
@1084588230 -zelazny.freenode.net- CIA-1 (micah@hyperreal.info) joined the channel
@1084590104 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084590476 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1084590486 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1084590768 <CIA-1!micah@hyperreal.info> commit by housel to fundev: Bug: 6995
@1084590768 <CIA-1!micah@hyperreal.info> Fix and improve the integration with the MPS garbage collector.
@1084590768 <CIA-1!micah@hyperreal.info> * collector.c
@1084590768 <CIA-1!micah@hyperreal.info> (primitive_copy_r): Stop violating the allocator contract by
@1084590768 <CIA-1!micah@hyperreal.info> <...>
@1084591788 -dircproxy- You disconnected
@1084597448 -dircproxy- You connected
@1084598928 <CIA-1!micah@hyperreal.info> commit by housel to fundev: Bug: 7012
@1084598929 <CIA-1!micah@hyperreal.info> Don't count both the intermediate warning count and the final warnings/errors
@1084598929 <CIA-1!micah@hyperreal.info> count output lines from the compiler.
@1084599440 -dircproxy- You disconnected
@1084600152 -dircproxy- You connected
@1084604524 <CIA-1!micah@hyperreal.info> commit by housel to fundev: Bug: 7007
@1084604525 <CIA-1!micah@hyperreal.info> Restore this file, which the Win32 release scripts seem to think they
@1084604525 <CIA-1!micah@hyperreal.info> need.  I have no idea where this DLL came from or what it does, though.
@1084606152 -zelazny.freenode.net- iehon (~char@vsat-148-63-65-106.c002.t7.mrt.starband.net) left the channel
@1084610764 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084610765 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084614463 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084614493 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If I fix more than one bug with a single commit, how do I indicate all the bugs in the check-in comment?
@1084614991 <CIA-1!micah@hyperreal.info> commit by cpage to src: Bug: 7013
@1084614991 <CIA-1!micah@hyperreal.info> Added support for casting unsigned values to signed integer types. Melange didn't used to differentiate signed and unsigned types. Note that Melange may still need additional code to handle casting from signed to unsigned.
@1084615580 <CIA-1!micah@hyperreal.info> commit by cpage to src: Bug: 7014 7015
@1084615580 <CIA-1!micah@hyperreal.info> Fixed parsing of backslash line continuations and character
@1084615580 <CIA-1!micah@hyperreal.info> literals. Backslashes used to indicate line continuation were only
@1084615580 <CIA-1!micah@hyperreal.info> handled on preprocessor lines. Character literals were greedy and
@1084615580 <CIA-1!micah@hyperreal.info> <...>
@1084615869 -zelazny.freenode.net- gabor (root@cantor.gwydiondylan.org) joined the channel
@1084615901 <gabor!root@cantor.gwydiondylan.org> re.
@1084617514 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084620313 <gabor!root@cantor.gwydiondylan.org> hi andreas!
@1084620331 [gabor!root@cantor.gwydiondylan.org] ACTION just had lunch
@1084621194 <andreas!andreas@cantor.gwydiondylan.org> Hi Gabor! I'm about to prepare lunch, it's asparagus with sauce hollandaise today.
@1084621276 -zelazny.freenode.net- brucehoult (~bruce@202-0-60-44.adsl.paradise.net.nz) joined the channel
@1084621338 <andreas!andreas@cantor.gwydiondylan.org> Good evening Bruce!
@1084621354 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> hi Andreas!
@1084621362 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> Peter's stats are interesting.
@1084621373 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> makes Boehm look quite good, actually
@1084621377 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning *
@1084621387 <gabor!root@cantor.gwydiondylan.org> andreas: I had that thursday evening. today I had a pizza
@1084621407 <gabor!root@cantor.gwydiondylan.org> hi *
@1084621417 <andreas!andreas@cantor.gwydiondylan.org> Hi Hannes.
@1084621422 <gabor!root@cantor.gwydiondylan.org> which stats?
@1084621428 <andreas!andreas@cantor.gwydiondylan.org> What stats? I must have missed them...
@1084621434 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> http://www.gwydiondylan.org/~housel/gcbench
@1084621496 <andreas!andreas@cantor.gwydiondylan.org> The overhead I'm observing is *much* bigger, something like a factor of 10. I'm trying the hot MPS, to see whether that makes a difference.
@1084621525 <andreas!andreas@cantor.gwydiondylan.org> My guess is that the read/write barriers are horribly inefficient.
@1084621632 <andreas!andreas@cantor.gwydiondylan.org> Of course, it would be a fairer comparison if Boehm were oeprated in incremental mode too.
@1084621662 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> it's taking around 4 min more user time than Boehm, and receiving 10,000 signals for each second of that
@1084621699 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> I can easily believe 100 uS per signal
@1084621751 <andreas!andreas@cantor.gwydiondylan.org> Yes,. Time for some mprotects/munprotects and the appropriate context switches must be added to the signal delivery overhead too.
@1084621835 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> I have a hard time seeing incremental GC being of much use, except in *very* demanding realtime applications
@1084621856 <andreas!andreas@cantor.gwydiondylan.org> Like games, for instance.
@1084621879 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> when you have a CPU with 2 - 3 GB/sec of memory bandwidth, and programs are stil lonly in the tens or hundreds of MB ... what, really, is the problem with stop-the-world?
@1084621899 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> games.  pfft.  they're only ~60 Hz realtime
@1084621913 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> ok, 85 - 100 :-)
@1084621928 <andreas!andreas@cantor.gwydiondylan.org> I *did* notice the difference that turning on incremental mode made.
@1084621941 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> in what?
@1084621950 <andreas!andreas@cantor.gwydiondylan.org> In the VRML viewer.
@1084621950 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> in smoothness?
@1084621960 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084621968 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> did you try doing a forced gc() in the main loop?
@1084621974 <andreas!andreas@cantor.gwydiondylan.org> Nope.
@1084621986 <andreas!andreas@cantor.gwydiondylan.org> Might have been worth a try.
@1084621987 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> that's often all it takes
@1084622143 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> cpage: there is no mechanism for committing to more than one bug at a time
@1084622177 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> if the fix naturally fixes several bugs then commit to one of them and close the others as duplicates of the first one
@1084622295 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> I might be able to start working on d2c more soon, with a decent speed machine at home
@1084622342 <andreas!andreas@cantor.gwydiondylan.org> Good!
@1084622354 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> (AMD XP3200)
@1084622365 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> .. and a dual 2.0 G5 at work :-)
@1084622366 <andreas!andreas@cantor.gwydiondylan.org> "decent" is an euphemism then.
@1084622409 <andreas!andreas@cantor.gwydiondylan.org> Hm, even with the hot MPS, compiling the dylan library already took 2 minutes, and isn't finished yet. It was done with Boehm in under a minute.
@1084622467 <andreas!andreas@cantor.gwydiondylan.org> Ok, I'm off for cooking. Later!
@1084622481 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> I think I've had my current Athlon 700 for four years now.  A pretty good run.
@1084636550 <housel!housel@cantor.gwydiondylan.org> those stats were with hot mps
@1084636620 <housel!housel@cantor.gwydiondylan.org> I sent Richard Brooksby a question about tuning the generational gc... the values in there are a bit arbitrary
@1084637336 <andreas!andreas@cantor.gwydiondylan.org> Morning Peter! Good work with the GC, it doesn't crash anymore now.
@1084637391 <andreas!andreas@cantor.gwydiondylan.org> Although I still observe the speed problems. Programs are crawling to a halt after a while. It takes about an hour to build the Dylan library, compared to about a minute with Boehm.
@1084637672 <housel!housel@cantor.gwydiondylan.org> hmmm... obviously it didn't take that long for me
@1084637673 <andreas!andreas@cantor.gwydiondylan.org> An strace on the process contains the same few lines over and over again:
@1084637683 <andreas!andreas@cantor.gwydiondylan.org> --- SIGSEGV (Segmentation fault) @ 0 (0) ---
@1084637683 <andreas!andreas@cantor.gwydiondylan.org> mprotect(0x5dff2000, 28672, PROT_READ|PROT_WRITE|PROT_EXEC) = 0
@1084637683 <andreas!andreas@cantor.gwydiondylan.org> mprotect(0x5dff2000, 28672, PROT_READ|PROT_EXEC) = 0
@1084637683 <andreas!andreas@cantor.gwydiondylan.org> sigreturn()                             = ? (mask now [PIPE RTMIN])
@1084637705 <andreas!andreas@cantor.gwydiondylan.org> That in an endless loop. The addresses don't even vary.
@1084637719 <housel!housel@cantor.gwydiondylan.org> bleah
@1084637730 <andreas!andreas@cantor.gwydiondylan.org> Oh wait, it doesn't even ever finish. Seems to hand.
@1084637734 <andreas!andreas@cantor.gwydiondylan.org> hang even.
@1084637748 <housel!housel@cantor.gwydiondylan.org> Maybe you should try the trunk development MPS
@1084637764 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084637855 <andreas!andreas@cantor.gwydiondylan.org> Where do I get it from?
@1084637911 <housel!housel@cantor.gwydiondylan.org> http://www.ravenbrook.com/project/mps/master/code I think
@1084638061 <housel!housel@cantor.gwydiondylan.org> http://www.ravenbrook.com/project/mps/version/1.104/ is a release snapshot
@1084638061 <andreas!andreas@cantor.gwydiondylan.org> 'k. Maybe it has to do with the sigaction code you commented out, though. I've re-enabled it. I try to turn it off again.
@1084638612 <andreas!andreas@cantor.gwydiondylan.org> Do you happen to know what the wi and ti versions are?
@1084638632 <housel!housel@cantor.gwydiondylan.org> wi = white-hot
@1084638637 <housel!housel@cantor.gwydiondylan.org> ti = telemetry
@1084638812 <andreas!andreas@cantor.gwydiondylan.org> What's the difference between hot and white-hot?
@1084638825 <housel!housel@cantor.gwydiondylan.org> optimization flags I think
@1084638984 <andreas!andreas@cantor.gwydiondylan.org> BTW: the linux runtime stores the TEB pointer at (end-of-stack - 0xff). I really wonder whether that is safe.
@1084639066 <andreas!andreas@cantor.gwydiondylan.org> Ok, disabling the sigaction calls didn't do it. Now trying the trunk code...
@1084639359 <andreas!andreas@cantor.gwydiondylan.org> Ok, I'm confused. What is mmdw.a, and how is it built?
@1084639395 <housel!housel@cantor.gwydiondylan.org> mmdw.a is a non-default target
@1084639421 <andreas!andreas@cantor.gwydiondylan.org> I see.
@1084639422 <housel!housel@cantor.gwydiondylan.org> use "make -f lii4gc.gmk mmdw.a mpsliban.a"
@1084639450 <andreas!andreas@cantor.gwydiondylan.org> "mpsliban"?
@1084639493 <housel!housel@cantor.gwydiondylan.org> it's the plinth library
@1084639500 <andreas!andreas@cantor.gwydiondylan.org> mpsplan, for sure.
@1084639531 <housel!housel@cantor.gwydiondylan.org> right (I'm in win32, that was from memory)
@1084639544 <housel!housel@cantor.gwydiondylan.org> you need to copy both libs into pentium-linux
@1084639580 <andreas!andreas@cantor.gwydiondylan.org> None of the script files references either mps.a or mpsplan.a. You're sure they're actually needed?
@1084639613 <housel!housel@cantor.gwydiondylan.org> mps.a isn't needed
@1084639634 <housel!housel@cantor.gwydiondylan.org> I can't remember how the plinth gets included but I'm sure it does
@1084639667 <andreas!andreas@cantor.gwydiondylan.org> What's a plinth?
@1084639897 <housel!housel@cantor.gwydiondylan.org> low-level platform layer
@1084639984 <housel!housel@cantor.gwydiondylan.org> I/O, mostly
@1084640495 <andreas!andreas@cantor.gwydiondylan.org> *sigh* Same problem again. Removing all the Bootstrap directories and trying again...
@1084640600 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084640642 <housel!housel@cantor.gwydiondylan.org> but cold works?
@1084640700 <andreas!andreas@cantor.gwydiondylan.org> It used to work this morning, and now it doesn't. I simply suspect I've mixed up some .a files.
@1084641164 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084641202 <housel!housel@cantor.gwydiondylan.org> the MPS page fault stuff is pretty nifty... it actually disassembles the faulting instruction and executes it
@1084641261 <andreas!andreas@cantor.gwydiondylan.org> Oh, now *that*s something I'd want to port to a new architecture. :)
@1084641311 <andreas!andreas@cantor.gwydiondylan.org> But nifty indeed.
@1084641329 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084641330 <andreas!andreas@cantor.gwydiondylan.org> Probably beats makring the whole page as grey again.
@1084641336 <andreas!andreas@cantor.gwydiondylan.org> Hi Tim!
@1084641476 <andreas!andreas@cantor.gwydiondylan.org> ARGH! The problem persists!
@1084641505 <housel!housel@cantor.gwydiondylan.org> that's weird
@1084641523 <housel!housel@cantor.gwydiondylan.org> try building the MPS tests and see if they work
@1084641541 <andreas!andreas@cantor.gwydiondylan.org> Ok. What target?
@1084641553 [housel!housel@cantor.gwydiondylan.org] ACTION checks
@1084641642 <housel!housel@cantor.gwydiondylan.org> look in comm.gmk iin the section labeled "genuine targets"
@1084641650 <housel!housel@cantor.gwydiondylan.org> all of those are tests
@1084641665 <housel!housel@cantor.gwydiondylan.org> oh... try target "testrun"
@1084642053 <andreas!andreas@cantor.gwydiondylan.org> Ok, after exporting VARIETY tests run. Look good so far.
@1084642118 -zelazny.freenode.net- hannes (~hannes@pD9E7B555.dip.t-dialin.net) joined the channel
@1084642119 <andreas!andreas@cantor.gwydiondylan.org> Conclusion:  Failed to find any defects.
@1084642121 <hannes!~hannes@pD9E7B555.dip.t-dialin.net> re
@1084642145 <andreas!andreas@cantor.gwydiondylan.org> Hi Hannes. MPS on Linux works for Peter, did work for me this morning.
@1084642165 <hannes!~hannes@pD9E7B555.dip.t-dialin.net> andreas: great, also with multiple threads?
@1084642174 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084642199 <andreas!andreas@cantor.gwydiondylan.org> I *did* mention it stopped working for me for no apparent reason?
@1084642209 <hannes!~hannes@pD9E7B555.dip.t-dialin.net> housel: with linux emulation on FreeBSD or with a "real" linux?
@1084642214 <housel!housel@cantor.gwydiondylan.org> you might want to try running "cvspurge" in lib/run-time
@1084642249 <housel!housel@cantor.gwydiondylan.org> I'm using linux emulation, but I had to hack MPS because the linuxulator gets page fault sigcontext wrong
@1084642252 <andreas!andreas@cantor.gwydiondylan.org> I have no such command.
@1084642261 <housel!housel@cantor.gwydiondylan.org> it's in the cvsutils package
@1084642304 <hannes!~hannes@pD9E7B555.dip.t-dialin.net> housel: i read your mail to freebsd-hackers ;)
@1084642385 <housel!housel@cantor.gwydiondylan.org> no responses though, sigh
@1084643348 <andreas!andreas@cantor.gwydiondylan.org> Ok. This morning, I used a Boehm stage 2 bootstrap to build a stage 3, and used the result of that to build gctest. Later, I tried to build an MPS stage 2, and this results in a non-working compiler.
@1084643399 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084643399 <andreas!andreas@cantor.gwydiondylan.org> Now, I've tried to build gctest in stage 2. This sort of works, the master threads does its work, but all the other threads hang, waiting for a lock in a hash table.
@1084643654 <andreas!andreas@cantor.gwydiondylan.org> With just a single thread, gctest works fine, including weak links.
@1084643663 <housel!housel@cantor.gwydiondylan.org> I'll try gctest in win32 as soon as this build finishes
@1084643736 <housel!housel@cantor.gwydiondylan.org> the compiler doesn't use weak tables BTW
@1084643826 <andreas!andreas@cantor.gwydiondylan.org> With one master thread and one spawned thread, I get a looping garbage collector, again mprotecting the same address as seen before over and over again.
@1084644515 <andreas!andreas@cantor.gwydiondylan.org> Hmmmm. There's code from x86-linux-exceptions.c being called. Maybe I try disabling more there.
@1084644531 <andreas!andreas@cantor.gwydiondylan.org> It keeps segfaulting on the same address over and over again, btw.
@1084644591 <housel!housel@cantor.gwydiondylan.org> what's at that address?
@1084644600 <housel!housel@cantor.gwydiondylan.org> is it really one of MPS's protected pages?
@1084644624 <andreas!andreas@cantor.gwydiondylan.org> It looks like it's some valid data, yes.
@1084644773 <andreas!andreas@cantor.gwydiondylan.org> If I dereference the address three times, I end up at KLsimple_object_vectorGVKdW.
@1084644802 <housel!housel@cantor.gwydiondylan.org> ok
@1084644830 <andreas!andreas@cantor.gwydiondylan.org> You said that MPS actually disassembles the faulting instruction?
@1084644837 <housel!housel@cantor.gwydiondylan.org> yes
@1084644849 <housel!housel@cantor.gwydiondylan.org> proti3.c IIRC
@1084644857 <andreas!andreas@cantor.gwydiondylan.org> So its usual mode of operation is returning to the instruction after the faulting one?
@1084644879 <housel!housel@cantor.gwydiondylan.org> I think so
@1084644888 <housel!housel@cantor.gwydiondylan.org> if it's a move instruction it recognizes
@1084644899 <andreas!andreas@cantor.gwydiondylan.org> Well, it keeps returning to the faulting address.
@1084644912 <andreas!andreas@cantor.gwydiondylan.org> 0x40368fe2 <Ktry_to_puthash_newVKiI+312>:       mov    %ecx,0xb(%edi,%esi,1)
@1084644951 <housel!housel@cantor.gwydiondylan.org> hmmm... I think it should be able to recognize that
@1084644988 <andreas!andreas@cantor.gwydiondylan.org> What mechanism does it use to set the address at which the code is supposed to continue?
@1084644994 <housel!housel@cantor.gwydiondylan.org> interesting that I remember seeing that very instruction at one point
@1084645011 <housel!housel@cantor.gwydiondylan.org> it munges the sigcontext and calls sigreturn
@1084645060 <andreas!andreas@cantor.gwydiondylan.org> Which is highly non-POSIXish.
@1084645106 <housel!housel@cantor.gwydiondylan.org> which is why the file is called protlii3.c
@1084645129 [andreas!andreas@cantor.gwydiondylan.org] ACTION pulls that into the editor.
@1084645797 <andreas!andreas@cantor.gwydiondylan.org> Ok, I'm removing all traces of signaÃl handling from the runtime.
@1084645972 <housel!housel@cantor.gwydiondylan.org> you don't have to rebuild everything if you rebuild the runtime... just kill bin/dylan, lib/libdylan.so, and Build/dylan/dylan.so
@1084645987 <housel!housel@cantor.gwydiondylan.org> saves time
@1084646002 <andreas!andreas@cantor.gwydiondylan.org> Ok.
@1084646619 <andreas!andreas@cantor.gwydiondylan.org> Ok. Now gctest with two threads works. Again I see the pattern that the master thread is quite fast, and the second thread crawls.
@1084646625 <housel!housel@cantor.gwydiondylan.org> heh... I just got an error under win32 at: mov edi,dword ptr [ebx+edi+0Bh]
@1084646646 <andreas!andreas@cantor.gwydiondylan.org> Weird.
@1084646656 <andreas!andreas@cantor.gwydiondylan.org> Bootstrap phase 2 proceeds.
@1084646848 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1084648120 <andreas!andreas@cantor.gwydiondylan.org> Still waiting for stage 2. I have a serious performance problem in the GC here.
@1084648786 <andreas!andreas@cantor.gwydiondylan.org> Building linker-support... 
@1084648787 <andreas!andreas@cantor.gwydiondylan.org> MPS ASSERTION FAILURE: res == 0
@1084648787 <andreas!andreas@cantor.gwydiondylan.org> lockli.c
@1084648787 <andreas!andreas@cantor.gwydiondylan.org> 140
@1084648791 <andreas!andreas@cantor.gwydiondylan.org> *sigh*
@1084648815 <housel!housel@cantor.gwydiondylan.org> that's something, I guess
@1084648859 <housel!housel@cantor.gwydiondylan.org> *that* is something that isn't supposed to happen
@1084648865 <housel!housel@cantor.gwydiondylan.org> stack trace?
@1084649006 <andreas!andreas@cantor.gwydiondylan.org> Hmmm... need to tweak env vars to run the compiler in gdb...
@1084649419 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084649424 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "cpage: there is no mechanism for committing to more than one bug at a time"
@1084649432 <housel!housel@cantor.gwydiondylan.org> hi chris
@1084649437 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, there is: I took a guess and just put the numbers on the same line and it worked.
@1084649453 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Bug: 1234 5321"
@1084649462 <housel!housel@cantor.gwydiondylan.org> nifty
@1084649486 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> My first guess was with a comma, and it refused to accept the commit. That's nice.
@1084649499 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Nice that it simply refused the commit in the case of an error.)
@1084649571 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> IIRC, MPS required some tuning for HD. There's a setting for the amount of memory it's allowed to allocate before it starts collecting. I remember we had to adjust that number higher and higher over time.
@1084649595 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...to balance between avoiding collections and VM paging.
@1084649654 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Also, I'm not entirely sure what did and didn't get implemented, but we talked about several different strategies for allocation, and the compiler was supposed to choose from among them using information it had about the program.
@1084649676 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> One obvious one is stack-based allocation.
@1084649748 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> As a general comment, unless you can guarantee GC will take less than a tenth of a second, incremental is vital for any interactive application that has to respond to user input. That includes HD itself, which never suffered user-detectable delays.
@1084649752 <andreas!andreas@cantor.gwydiondylan.org> Hi Chris! For now, I could live with a less-than-optimal performance, if it worked at all without crashing.
@1084649825 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Crashing = Infinitely slow performance
@1084649834 <housel!housel@cantor.gwydiondylan.org> :-)
@1084649898 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Does MPS actually require a signal to do incremental GC? I thought the strategy was to just run the collector a little bit every time the program entered the allocator?
@1084649914 <andreas!andreas@cantor.gwydiondylan.org> It needs signals to implement read and write barriers.
@1084650085 <housel!housel@cantor.gwydiondylan.org> andreas: is your collector.c exactly what is in the repos now, or does it contain some of the changes from my patch yesterday?
@1084650091 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why are barriers involved in the performance test under discussion? What needs to be blocked from reading/writing?
@1084650119 <housel!housel@cantor.gwydiondylan.org> read section 6.3 of the paper
@1084650190 <housel!housel@cantor.gwydiondylan.org> basically to do scanning incrementally you need to make sure work you've already done hasn't been invalidated by the mutator
@1084650193 <andreas!andreas@cantor.gwydiondylan.org> I still have the asserts in there, the rest is as in the CVS.
@1084650215 <housel!housel@cantor.gwydiondylan.org> kill those, they're unnecessary
@1084650227 <housel!housel@cantor.gwydiondylan.org> that's where your performance went
@1084650235 <andreas!andreas@cantor.gwydiondylan.org> I hope so.
@1084650288 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> So discourage mutation!
@1084650303 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> +1
@1084650306 <andreas!andreas@cantor.gwydiondylan.org> Certainly! Hi Riastradh.
@1084650349 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Hi.
@1084650366 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If that's really the performance killer, it sounds like either there's too much mutation or the mutation patterns in this case are at odds with the collector patterns.
@1084650383 <andreas!andreas@cantor.gwydiondylan.org> BTW: there are read barriers too...
@1084650396 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is there a way to tune MPS to reduce the number of signals, I wonder?
@1084650413 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> There's probably too much mutation.  In general, when reading & writing Dylan code (well, to the extent that I write it), I find myself seeing & using much more mutation than with Scheme.
@1084650523 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm a little puzzled, though: It's also generational, right? Shouldn't the number of signals go down once it's moved a lot of the mutating stuff into an older generation?
@1084650578 <andreas!andreas@cantor.gwydiondylan.org> Well, once the code works, we can play with the telemetry stuff and generate nice graphs.
@1084650624 <andreas!andreas@cantor.gwydiondylan.org> But no, this won't help you with read barriers at all.
@1084650688 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, there has to be a way to tune it to reduce barrier signalling, one way or another. That things got more options than a new car.
@1084650719 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I assume there's also a way to get it to be non-incremental, if desirable.
@1084650936 <housel!housel@cantor.gwydiondylan.org> that's not clear from the paper
@1084650967 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION is looking at the paper
@1084651519 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> hey all
@1084651540 <andreas!andreas@cantor.gwydiondylan.org> Hi!
@1084651553 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Harlequin's DylanWorks?" Heh. I think that name was considered, but was never actually used for Harlequin Dylan.
@1084651559 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> cpage: Excellent comment about crashing = infinitely slow
@1084651565 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> more C++ programmers should understand that
@1084651587 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> minimal-pentium-dw?
@1084651597 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> DW = DylanWorks
@1084651699 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I mean the name "DylanWorks" was never used for the shipping product.
@1084651738 <housel!housel@cantor.gwydiondylan.org> why not?
@1084651769 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I don't know that there's an answer to that, other than we went with "Harlequin Dylan".
@1084651829 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> IMO, I think "LispWorks" sounds fine, but "DylanWorks" has one-too-many syllables.
@1084651851 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But I didn't choose the name, in any case.
@1084651991 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And I certainly didn't choose "Liquid Common Lisp". :-)
@1084652096 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> DylanWorks has a ring to it, but so does  Harlequin Dylan
@1084652104 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Hmm. The MPS paper has links to Dylan and ScriptWorks requirements, but they're dead.
@1084652122 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> Harlequin is an awesome name
@1084652142 <housel!housel@cantor.gwydiondylan.org> yes... I wrote a letter to richard brooksby asking him to put the dylan requirements online
@1084652148 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, it probably wouldn't just be "DylanWorks," it would be, "Harlequin DylanWorks," which is quite a mouthful.
@1084652165 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> true
@1084652188 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> Microsoft Office System? (7 syllables)
@1084652237 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> my, aren't I random today
@1084652275 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "MSN Messenger for Mac version 4.0"
@1084652297 <housel!housel@cantor.gwydiondylan.org> Microsoft Windows for Workgroups version 3.11
@1084652320 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...Service Pack 3 Release 2
@1084652395 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Looks like MS has actually shortened some names. e.g., "Office v. X"
@1084652626 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Section 5.1 is interesting. Describes the types of pools HD used.
@1084652658 <housel!housel@cantor.gwydiondylan.org> uses
@1084652664 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> +1
@1084652671 <housel!housel@cantor.gwydiondylan.org> AMC, AWL, and LO
@1084652750 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The "A" pool isn't in the sources? Too bad. That'd be a handy pool for some things.
@1084652752 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> cpage: Are you keeping score :)
@1084652770 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> is the API to MPS exposed in FD?
@1084652793 <housel!housel@cantor.gwydiondylan.org> some of it, through primitive-* functions
@1084652806 <andreas!andreas@cantor.gwydiondylan.org> There's a memory-manager library. Never looked into it.
@1084652864 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> rhashem: The total will be announced at Milliways.
@1084652906 <housel!housel@cantor.gwydiondylan.org> I just looked at it... it calls primitive-* functions
@1084652984 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is primitive-* a compiler primitive?
@1084652995 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> milliways?
@1084652997 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/is/are
@1084653005 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm glad you asked: <http://www.bbc.co.uk/cult/hitchhikers/guide/milliways.shtml>
@1084653064 <housel!housel@cantor.gwydiondylan.org> yes... the compiler emits HARP instructions for them
@1084653088 <housel!housel@cantor.gwydiondylan.org> most of these particular HARP instrucitons are implemented as C calls
@1084653090 <andreas!andreas@cantor.gwydiondylan.org> Actually, some are implemented in C.
@1084653098 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> cpage: heh
@1084653164 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> in regards to the recent discussion on the dylan mailing list about alternate backends --- it seems to be that having full control over the code generator as you do with HARP is very useful
@1084653242 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Now *that's* an understatement. ;)
@1084653248 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> I don't see other back-ends having that same degree of control
@1084653866 <housel!housel@cantor.gwydiondylan.org> do any of you guys have any opinions on bug 6986?
@1084653980 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Which part? That's a pretty long bug report.
@1084653992 <housel!housel@cantor.gwydiondylan.org> all of it
@1084654032 <housel!housel@cantor.gwydiondylan.org> mostly about which edition restrictions to remove
@1084654066 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> [Reading through it.] I'd change the Register button (the name and the action) to just go to GwydionDylan.org.
@1084654078 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...until we have some other use for it.
@1084654143 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm not fond of the About Box.
@1084654154 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If I get time I'll propose something for it.
@1084654223 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why have editions any more?
@1084654326 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think we still need to distinguish "Internal" as a version for developing the environment itself.
@1084654386 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I wouldn't strip out the predicates wholesale (why throw away information?), but I'd reduce it to a single edition.
@1084654401 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Dispatch color ".el" files are written in the build directory if release-internal?() is true."
@1084654423 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is that separate from the dispatch coloring in the IDE or is it the same thing?
@1084654451 <andreas!andreas@cantor.gwydiondylan.org> I think it's for dispatch coloring in emacs?
@1084654460 <housel!housel@cantor.gwydiondylan.org> yes
@1084654500 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If that works, I say turn it into a user preference and make it available in non-internal releases.
@1084654559 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> On the other hand, it's important to make the IDE stand on it's own, and seduce people away from other tools. :-)
@1084654582 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But it sounds like something worth having in the command-line compiler.
@1084654600 <andreas!andreas@cantor.gwydiondylan.org> I think the plan was turning Deuce into a full-blown Emacs replacement. :)
@1084654644 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> For editing code, sure, although I hope "full-blown" doesn't actually mean "full-blown" :-O
@1084654659 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> eDylan, anyone?
@1084654669 <housel!housel@cantor.gwydiondylan.org> I confess that I still use Emacs even when using the IDE
@1084654689 <andreas!andreas@cantor.gwydiondylan.org> Why should the extension languae be any different from the implementation language? See Multics Emacs.
@1084654716 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> No, I just meant that code editing is only about 10% of emacs.
@1084654741 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Maybe less.)
@1084654772 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I *would* like to see the IDE support plugins and/or "scripts" written in various languages, including an interpreted or JIT-compiled Dylan.
@1084654788 <andreas!andreas@cantor.gwydiondylan.org> Now the ultra secret master plan of course is eliminating the need for anything written in C, inclduing UNIX/Windows. In that respect, it is like emacs.
@1084654811 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Shhhh! You'll blow our cover.
@1084654837 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ...um...don't you, uh...compile to C?
@1084654846 <andreas!andreas@cantor.gwydiondylan.org> Not with FD.
@1084654847 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That's only temporary.
@1084654864 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Of course what you want to do is compile to low-level Dylan. :)
@1084654866 <andreas!andreas@cantor.gwydiondylan.org> "written in" != "compiled to".
@1084654996 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> What will the run-time system, garbage collector, et cetera be written in?
@1084655025 <housel!housel@cantor.gwydiondylan.org> FD has a run-time generator
@1084655064 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 'Run-time generator?'
@1084655087 <housel!housel@cantor.gwydiondylan.org> input = definitions for run-time operations and utility functions
@1084655099 <housel!housel@cantor.gwydiondylan.org> output = dylan-support.o
@1084655100 <andreas!andreas@cantor.gwydiondylan.org> et cetera will be written in Dylan. The garbage collector *could* be written in Dylan, although I think MPS will serve us quite some time.
@1084655500 <housel!housel@cantor.gwydiondylan.org> back to editions... I think we need to have two editions, but it's hard to decide what to call them
@1084655801 <andreas!andreas@cantor.gwydiondylan.org> Hm, you're sure one edition isn't enough?
@1084656081 <housel!housel@cantor.gwydiondylan.org> internal at least would preserve #4, #7, #8, #9, #10, #11, #14, #15
@1084656125 <housel!housel@cantor.gwydiondylan.org> The other distinctions can go away
@1084656420 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What is #4?
@1084656480 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is #7 something useful outside of developing the IDE? Should it be a user preference instead?
@1084656495 <housel!housel@cantor.gwydiondylan.org> I think (haven't researched it) that's what causes warning messages to go to build/library-name.log instead of *standard-output*
@1084656519 -zelazny.freenode.net- CIA-5 (micah@hyperreal.info) joined the channel
@1084656574 <housel!housel@cantor.gwydiondylan.org> #7 could be a user-preference I guess
@1084656583 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> #9 definitely looks to me like either a user preference or it should depend on whether or not you have the sources around.
@1084656613 <andreas!andreas@cantor.gwydiondylan.org> You'll have the sources around. :)
@1084656645 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The real question is why do the system projects produce warnings at all? They should probably be addressed one way (fixing them) or another (suppressing them to indicate that they're acceptable.)
@1084656677 <housel!housel@cantor.gwydiondylan.org> that's a bit harder to do
@1084656720 <andreas!andreas@cantor.gwydiondylan.org> Why should there be a difference between system and user projects?
@1084656740 <housel!housel@cantor.gwydiondylan.org> Even though I do have sources around, I don't always want to know that Dylan has 18 warnings and 4 serious warnings
@1084656781 <andreas!andreas@cantor.gwydiondylan.org> That goes for *all* libraries, doesn't it?
@1084656796 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> andreas: I agree in general, but it depends. If the "system" projects are used everywhere, and using them produces a lot of warnings, I can imagine wanting to preserve the option to suppress them.
@1084656826 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yes, andreas. I'm just saying for now let's leave the existing behavior in place and worry about solving the more general problem later.
@1084656873 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> IMO, though, the longer-term solution is to resolve all warnings.
@1084656905 -zelazny.freenode.net- CIA-5 (micah@hyperreal.info) joined the channel
@1084657023 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think #10 might make more sense as a context-dependent behavior: If the environment itself is attached to a debugger, go directly to the debugger, else let serious condition handlers do their thing.
@1084657078 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, I think that should just be a debugger option (e.g., "Stop on serious conditions"), if possible.
@1084657182 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> #14 sounds like another user preference (to be specific, a "View" setting that can be changed in a lightweight manner, rather than literally in the preferences window.)
@1084657232 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> #15 sounds like something that needs to be fixed before it's ready for the release edition.
@1084657275 <housel!housel@cantor.gwydiondylan.org> ... if we only understood exactly what it meant
@1084657293 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, yeah, there's that, too.
@1084657309 <housel!housel@cantor.gwydiondylan.org> I imagine preserving object identity is hard to do in the presence of a relocating gc
@1084657317 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is #22 just commands that help developing the IDE itself?
@1084657355 <housel!housel@cantor.gwydiondylan.org> the only one I've ever used is "Open Registry Project"
@1084657363 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What's that do?
@1084657391 <housel!housel@cantor.gwydiondylan.org> opens a project via the Sources/registry entries (rather than a .lid/.hdp)
@1084657403 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION is getting more anxious about buying a PC or VirtualPC so he can use FunDev
@1084657450 <housel!housel@cantor.gwydiondylan.org> heh... just buy me a 12" iMac and you'll get FunDev for mac :-)
@1084657551 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I might be able to arrange to loan you one.
@1084657611 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Also, just let me know what needs work and I'll contribute. I was waiting because I was under the impression that we hadn't got it bootstrapping easily yet.
@1084657688 <housel!housel@cantor.gwydiondylan.org> that's true enough... I (at least) haven't researched how to do cross-builds yet
@1084657756 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> For Mac support, what tasks are there? Does it first need a PowerPC backend?
@1084657782 <housel!housel@cantor.gwydiondylan.org> it has one, but it looks like the details of low-level thread-local storage have been left undone
@1084657790 <housel!housel@cantor.gwydiondylan.org> (or one could use the C backend)
@1084657795 <andreas!andreas@cantor.gwydiondylan.org> Should be doable, though.
@1084657828 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Okay, so if there's a workable backend, what then? Build system? Libraries?
@1084657881 <andreas!andreas@cantor.gwydiondylan.org> An MPS port? Maybe "just" making it all work together?
@1084657893 <housel!housel@cantor.gwydiondylan.org> yes, I think that's about right
@1084657929 <housel!housel@cantor.gwydiondylan.org> a matter of hacking at it until it works
@1084658007 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Does it have to be cross-targeted from Windows?
@1084658021 <andreas!andreas@cantor.gwydiondylan.org> Or from Linux.
@1084658028 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Is the IDE working on Linux?
@1084658045 <andreas!andreas@cantor.gwydiondylan.org> Nope. Neither is building the compiler from the IDE working. :)
@1084658066 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, I don't think we ever got the IDE to build the compiler.
@1084658111 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I assume what was missing was for the IDE build system to support more actions than just "compile" and "link".
@1084658155 <housel!housel@cantor.gwydiondylan.org> maybe so
@1084658183 <housel!housel@cantor.gwydiondylan.org> the Jam build-system may fix some of that
@1084658719 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION is off to see Gojira
@1084658803 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> <http://www.thecastrotheatre.com/p-list.html#godzilla>
@1084658809 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Enjoy the weekend.
@1084658822 <housel!housel@cantor.gwydiondylan.org> cul
@1084658824 <andreas!andreas@cantor.gwydiondylan.org> Bye Chris!
@1084658910 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> Chris: Hey it *did* work, you're right!  I did not know that.
@1084658920 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What worked?
@1084658925 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> the commit message is on both bugs
@1084658929 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah.
@1084658943 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> 7014/7015
@1084658978 <brucehoult!~bruce@202-0-60-44.adsl.paradise.net.nz> and: http://www.gwydiondylan.org/cgi-bin/cvszilla/transaction.cgi?id=68
@1084661380 -zelazny.freenode.net- hannes_ (~hannes@pD9E7AC0E.dip.t-dialin.net) joined the channel
@1084675422 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084676259 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084680102 -zelazny.freenode.net- brucehoult (~bruce@202-0-60-44.adsl.paradise.net.nz) joined the channel
@1084687836 -zelazny.freenode.net- brucehoult_ (~bruce@202-0-60-37.adsl.paradise.net.nz) joined the channel
@1084688381 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084691275 -zelazny.freenode.net- brucehoult (~bruce@202.0.60.37) joined the channel
@1084693674 -zelazny.freenode.net- IRCMonkey (~chatzilla@wg02.ucr.edu) joined the channel
@1084693686 -zelazny.freenode.net- IRCMonkey (~chatzilla@wg02.ucr.edu) left the channel
@1084712344 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084712349 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> heya
@1084714078 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084714519 <andreas!andreas@cantor.gwydiondylan.org> Hi!
@1084714581 <andreas!andreas@cantor.gwydiondylan.org> We have name server trouble again. I'm working on it, might need a reboot of a machine a few hundred km away, though.
@1084714719 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Ew.
@1084714739 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> So when do we start using Koala as our webs server? ;-)
@1084714795 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Andreas, were you doing some work on getting Dylan running atop the L4 kernel?
@1084717259 <andreas!andreas@cantor.gwydiondylan.org> Hi Rob! We'll use koala as soon as the garbage collector runs stable with threads.
@1084717291 <andreas!andreas@cantor.gwydiondylan.org> And I did some work with Dylan on L4, but didn't get around doing much with it in the last few months. I'm thinking about porting MPS to L4, though.
@1084718109 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Cool.
@1084718146 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I noticed ioL4, and it made me think about running language environments on hardware (like Lisp Machines). 
@1084718194 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Although I don't know how much Dylan would benefit from custome microcode (maybe for GF dispatch?)
@1084718312 <andreas!andreas@cantor.gwydiondylan.org> In fact, one could run Dylan on the bare hardware. The benefit of L4 is that it is already there, has a reasonable abstraction, and allows a Linux system to run in a sandbox next to the Dylan OS.
@1084718332 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I see.
@1084718376 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> reading up on Lisp Machines, I notided that code was always debuggable and editable for any process on the system, and that you could get a reference to any GUI widget in the editor by clicking on it.
@1084718397 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> notide -> noted | noticed . :-)
@1084718398 <andreas!andreas@cantor.gwydiondylan.org> That's how things are supposed to be. :)
@1084718439 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> That doesn't seem impossible with a Dylan system, particularly if there's a single DUIM server?
@1084718466 <andreas!andreas@cantor.gwydiondylan.org> Right. Dylan is semantically so close to Lisp that many of the concepts can be translated.
@1084718501 <andreas!andreas@cantor.gwydiondylan.org> One of the key insights is that the UNIXish concept of separate address spaces is the beginning of madness and confusion.
@1084718502 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But is it dynamic enough to replace running code modularly?
@1084718546 <andreas!andreas@cantor.gwydiondylan.org> The language definition doesn't prevent it, and FD has some support for replacing code in running applications.
@1084718559 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Cool.
@1084718608 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> UNIX spaces are quite nice when you've been used to MacOS's cross-process memory corruption, and you can bet that there'd be secuirty FUD over single-memory-space systems. But I suppose with a language that doesn't allow you to trash other people's objects with rogue pointers both cases are less of a problem.
@1084718637 <andreas!andreas@cantor.gwydiondylan.org> Right. Except that UNIX confuses the concept of address spaces and protection domains.
@1084718897 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION goes boss-eyed.
@1084718928 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hwo do you mean? Is there any other reason for having multiple address spaces other than hardware limitations or security?
@1084718993 <andreas!andreas@cantor.gwydiondylan.org> Yes: so every process can pretend it is alone on the system, starting at address zero, etc.
@1084719072 <andreas!andreas@cantor.gwydiondylan.org> In ancient times, there always was only one program running on a computer. Then, the concept of timesharing was invented, but all the old software was supposed to run unmodified.
@1084719110 <andreas!andreas@cantor.gwydiondylan.org> But you could have addresses valid across all processes in the system, and map/unmap pages depending on which process is active.
@1084719145 <andreas!andreas@cantor.gwydiondylan.org> So instead of writing lots of IPC code, you just pass pointers and call functions.
@1084719195 <housel!housel@cantor.gwydiondylan.org> there's also the issue of "each process wants 4GB of addressable space"
@1084719220 <andreas!andreas@cantor.gwydiondylan.org> 32 bit of address space is simply too small these days.
@1084719240 <housel!housel@cantor.gwydiondylan.org> yes
@1084721211 <andreas!andreas@cantor.gwydiondylan.org> Good news: with MPS 1.104 (as opposed to the trunk version) bootstrapping works.
@1084721334 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But can't you just use one ultra-large address space with a single object-graph root? If there's no dealing with memory and no way of grabbing pointers, you don't need separate address spaces, surely?
@1084721401 <andreas!andreas@cantor.gwydiondylan.org> For sure. Back when Clinton lived in the White House, he had a Lispm for a web server. It was never rooted, despite having a single address space.
@1084721510 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Well, that and there being very few people who'd know how to hack it. ;-)
@1084721643 <andreas!andreas@cantor.gwydiondylan.org> Well, the only weakness yo tend to find on Lisp systems is programmers using READ to communicate with the outside world, and forgetting to turn off the special 'eval' mechanism that allows to execute arbitrary Lisp code.
@1084721651 <andreas!andreas@cantor.gwydiondylan.org> andreas@unuk:~/fd-build/bin$ ./dsp-basics 
@1084721651 <andreas!andreas@cantor.gwydiondylan.org> dsp-basics: fmtdy.c:141: dylan_wrapper_check: Assertion `version == 2' failed.
@1084721651 <andreas!andreas@cantor.gwydiondylan.org> Aborted
@1084721727 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hey, this would be a plus for a Dylan-based system. No 'eval'. :-)
@1084721751 <andreas!andreas@cantor.gwydiondylan.org> Actually, we need eval sooner or later for an interactive system.
@1084721769 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> How so?
@1084721770 <andreas!andreas@cantor.gwydiondylan.org> It's just not part of the core language, but it can be a library.
@1084721784 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Don't we just compile on the fly?
@1084721799 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I mean, I know REPLs can compile in Lisp.
@1084721806 <andreas!andreas@cantor.gwydiondylan.org> That's an implementation detail. Conceptually, it's still eval.
@1084721839 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> But can we sandbox a Dylan REPL or make its effects transient. Or even read-only?
@1084721896 <andreas!andreas@cantor.gwydiondylan.org> I think that the sealing concepts can be applied here.
@1084721954 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> That's a good idea. Indeed a Dylan Machine could seal particular processes and sections of the file system.
@1084722004 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> As well as system calls, so you couldn't spoof them.
@1084722029 <andreas!andreas@cantor.gwydiondylan.org> For certain kinds of protection mechanisms, the concept of capabilities can be applied. Simply spoken: if you cannot get hold of a reference to an object, you can't reach it.
@1084723222 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> And since Methods are objects this works well in Dylan. But you need to make sure that people don't pass objects by IPC, circumventing such security. There's a Java attack where you can get a reference to an object by adding a "reference to previous object" bytecode to a serialisation stream and then reading it back (or something like that). It's quite hard to think of all the eventualities for a sufficeintly powerful environ
@1084723344 <andreas!andreas@cantor.gwydiondylan.org> Serializing byte code sounds like inviting trouble.
@1084723553 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Well, using Java is inviting trouble, but I digress... :-)
@1084723720 <andreas!andreas@cantor.gwydiondylan.org> At least they have the problem of buffer overflows under control.
@1084723896 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> True.
@1084724661 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Now all we need is sql-inject-lint...
@1084724753 <andreas!andreas@cantor.gwydiondylan.org> IIRC, the Dylan ODBC API makes it very hard to produce SQL injection bugs.
@1084724926 -zelazny.freenode.net- CIA-5 (micah@hyperreal.info) joined the channel
@1084725036 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> That would be good. We had an injection bug at work this week (in an ASP). I'd much rather be using Dylan.
@1084726107 <housel!housel@cantor.gwydiondylan.org> Are there still DNS problems?
@1084726230 <andreas!andreas@cantor.gwydiondylan.org> Yes. And I'm afraid they will continue until I can call the housing privoder during work hours.
@1084726296 <andreas!andreas@cantor.gwydiondylan.org> rob: I think koala on Linux will work within weeks, if not days. On Windows, it might be usable even today.
@1084726357 <housel!housel@cantor.gwydiondylan.org> do you have a backtrace for the assertion failure you just pasted?
@1084726383 <lisppaste!~lisppaste@69.55.225.112> andreas pasted "assertion failure in koala" at http://www.common-lisp.net/paste/display/967
@1084726440 <lisppaste!~lisppaste@69.55.225.112> andreas annotated #967 with "walking up the stack to show inlined functions too" at http://www.common-lisp.net/paste/display/967#1
@1084726499 <housel!housel@cantor.gwydiondylan.org> what does it think "version" is in frame #6?
@1084726538 <andreas!andreas@cantor.gwydiondylan.org> 0.
@1084726547 <andreas!andreas@cantor.gwydiondylan.org> What are wrappers, and how are they initialized?
@1084726564 <housel!housel@cantor.gwydiondylan.org> wrappers tell the gc about the layout of the class
@1084726576 <housel!housel@cantor.gwydiondylan.org> They're what word 0 of every object points to
@1084726693 <housel!housel@cantor.gwydiondylan.org> hmmm... I added that call to dylan_wrapper_check... I wonder if it's valid?
@1084726724 <housel!housel@cantor.gwydiondylan.org> (the one at line 1258)
@1084726749 <housel!housel@cantor.gwydiondylan.org> try getting rid of it and see if anything bad happens
@1084726774 <andreas!andreas@cantor.gwydiondylan.org> That's what I did. Now I get the same problem with a check in line 198.
@1084726785 <andreas!andreas@cantor.gwydiondylan.org> Inside dylan_check, that is.
@1084726803 <housel!housel@cantor.gwydiondylan.org> ok
@1084726815 [housel!housel@cantor.gwydiondylan.org] ACTION goes to church... back in ~3 hours
@1084726830 <andreas!andreas@cantor.gwydiondylan.org> See you! (or not, maybe we go to the movies)
@1084727296 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Koala? Good.
@1084727310 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> What are you going to see? 
@1084727377 <andreas!andreas@cantor.gwydiondylan.org> Troy. Want to see Brad Pitt with 10kg extra muscles. :)
@1084727436 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heh. I think I'm going to let Evie take Tom to see that one. I have a lower Brad tolerance than her. :-)
@1084727464 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Brad Pitt is actually a good actor...it's just that rarely do any directors want him to act.
@1084727581 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> he was good in "Interview". And 12 monkeys. Which Bruce Willis was good in as well (I was shocked how good he was in Pulp Fiction). Now I just need to find a way of relating this back to Dylan. :-) 
@1084727618 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Hmmmm.
@1084727775 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Well...if we consider the strongly hyped languages of atrocious design to be like the virus that wiped out civilization in _The 12 Monkeys_, and civilization to be the software industry (at least, that of it that isn't a horrid mess)...
@1084727987 -zelazny.freenode.net- pritlove (~pritlove@212.222.128.123) joined the channel
@1084728380 <andreas!andreas@cantor.gwydiondylan.org> Riastradh: http://www.dw-world.de/english/0,3367,1446_A_1201704_1_A,00.html
@1084728507 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The answer is clear. Compulsory licensing and training for scriptkiddies.
@1084728672 <andreas!andreas@cantor.gwydiondylan.org> We do what we can. We'll get involved into the civil lawsuit, to make sure the judge honors the responsibility Microsoft has for the damage being done by the worm.
@1084728708 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> That's good. Have you been lobbying over Software Patents?
@1084728729 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Is there a *PowerPC* VM for MacOSX? Something like Virtual PC on Windows or Mac-On-Linux on LinuxPPC? Something I could use to play with kernels on MacosX with...
@1084728759 <andreas!andreas@cantor.gwydiondylan.org> Yes, we did. The main work was done by the FFII folks, but we supported them.
@1084728838 <andreas!andreas@cantor.gwydiondylan.org> What really worries me about the software patent issue is that the European Commission is free to ignore decisions by the European Parliament.
@1084729045 <andreas!andreas@cantor.gwydiondylan.org> Ok, cinema time. Later!
@1084729083 -zelazny.freenode.net- hannes (~hannes@pD9E7AC0E.dip.t-dialin.net) joined the channel
@1084730105 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 'The hacker hierarchy, composed of hackers, crackers, and script kiddies' -- what crack is that moron smoking?
@1084730137 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 'Hackers hack websites to bring attention to security holes' -- ...
@1084731228 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Standard journalistic nonsense. Hacker == cracker. Hacking == cracking.
@1084731266 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> But he even uses the term 'cracker' to be something entirely different.
@1084734587 -zelazny.freenode.net- CIA-5 (micah@hyperreal.info) joined the channel
@1084734710 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084734731 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> robmeyers: "Is there a *PowerPC* VM for MacOSX?"
@1084734741 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> You mean VirtualPC?
@1084734781 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The only problem is it doesn't work on G5's, but of course you want the fastest machine possible to make sure it runs decently.
@1084734836 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think it's important to stress that Dylan was designed for interactive development. Just look at Apple Dylan and FunDev.
@1084734860 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> But it was designed with a focus on separating the dev runtime from the target runtime.
@1084734943 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Re Separate address spaces: You definitely want them to allow larger virtual spaces. It's effectively a kind of "extended addressing" like on x86, giving each process it's own "window."
@1084735001 -zelazny.freenode.net- pritlove (~pritlove@212.222.128.123) joined the channel
@1084735088 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Also, even if you make it impossible to unwittingly write code with buffer overflows, opening yourself up to that kind of attack, there's a security issue unless you either disallow machine instructions that access arbitrary addresses or introduce a protection scheme that makes most addresses read-only or inaccessible.
@1084735148 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If you do the latter, then you have to change the set of accessible pages as you switch from one process/thread to another.
@1084735180 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If you do that, and "extended" address spaces are still useful, then I think you end up with a system pretty much like UNIX, anyway.
@1084735249 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The UNIX model allows sharing segments of addresses, so I don't see how having a single address space gives you something you can't get with the UNIX model.
@1084735359 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It seems to me it's only a slight difference in the details.
@1084740414 -zelazny.freenode.net- CIA-5 (micah@hyperreal.info) joined the channel
@1084740774 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1084742647 <prom!~prom@217.13.206.33> robmyers: there is a ppc system emulator in qemu (http://fabrice.bellard.free.fr/qemu/)
@1084742658 <prom!~prom@217.13.206.33> re... just for a moment.
@1084743446 <^self!~fn@66.209.66.105> how abotu pearpc?  http://pearpc.sf.net/
@1084743451 <^self!~fn@66.209.66.105> "about"
@1084743502 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I saw a piece on PearPC on TechTV, and they warned that it's still pretty buggy right now.
@1084743527 <^self!~fn@66.209.66.105> yeah, it's way incomplete.
@1084743541 <^self!~fn@66.209.66.105> doesn't even boot netbsd
@1084743546 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought the question was about emulators running on PowerPC.
@1084743560 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> prom: thanks
@1084743563 <^self!~fn@66.209.66.105> oh, oops.
@1084743589 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Yes, I need a window in macOSX I can play eith kernels in. :-)
@1084743595 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> eith -> with.
@1084743639 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> VirtualPC
@1084743653 <^self!~fn@66.209.66.105> offtopic: bochs is pretty good now, if you run it on linux/x86.  it does some virtualization.  so, bochs on my 1.4ghz athlon emulates a 200mhz box, instead of a 6mhz one.
@1084743773 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> VPC doesn't run PPC...
@1084743780 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> heya chris.
@1084743790 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought you wanted an x86 emulator.
@1084743799 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> No, PPC.
@1084743813 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I use VPC as an X86 emulator at work.
@1084744232 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> VirtualPC for Mac running VirtualPC for Windows.
@1084744244 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Be sure to use an overclocked Mac. ;-)
@1084744280 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Rather: VirtualPC for Mac running PearPC for Windows.
@1084745085 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> That's sick. ;-)
@1084745152 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I want to play with L4 and HURD but I don't want to reboot. I know, I'm lazy.
@1084745622 <bhoult!~bhoult@morannon.eservglobal.co.nz> were you saying there is something wrong with cantor?
@1084745646 <bhoult!~bhoult@morannon.eservglobal.co.nz> tinderbox is keeping it busy, but other than that it looks ok
@1084745686 <bhoult!~bhoult@morannon.eservglobal.co.nz> plenty of free swap, plenty of free disk.  No free CPU...
@1084745828 <bhoult!~bhoult@morannon.eservglobal.co.nz> it's not in the DNS, but is responding fine on 212.40.165.86
@1084746431 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Well, almost midnight local. Goodnight.
@1084746657 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Last night I couldn't get to gwydiondylan.org
@1084746688 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Make that: Right now...
@1084746692 <bhoult!~bhoult@morannon.eservglobal.co.nz> I cna't get to it now, by hostname
@1084746698 <bhoult!~bhoult@morannon.eservglobal.co.nz> but it works by IP
@1084746713 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What's the IP?
@1084746720 <bhoult!~bhoult@morannon.eservglobal.co.nz> 212.40.165.86
@1084746728 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084746757 <bhoult!~bhoult@morannon.eservglobal.co.nz> hi Andreas
@1084746766 <andreas!andreas@cantor.gwydiondylan.org> Hi Bruce!
@1084746793 <andreas!andreas@cantor.gwydiondylan.org> I'll have somebody reboot the name server tomorrow morning. 
@1084746887 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, Melange doesn't handle the case where a struct member name is the same as a typename. After a lot of experimentation, I'm unsure whether it's the grammar or the parser code that needs to change.
@1084746894 <andreas!andreas@cantor.gwydiondylan.org> BTW: The only excuse to go and see Troy is seeing Orlando Bloom and Brad Pitt naked. Other than that, it's only average entertainment...
@1084746908 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Any help with this would be appreciated.
@1084746939 <andreas!andreas@cantor.gwydiondylan.org> Chris: do you have some easy-to-reproduce sample code?
@1084746960 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net>   typedef int Foo; struct { char Foo; };
@1084747015 <andreas!andreas@cantor.gwydiondylan.org> Will look into it tomorrow. I'm about to go to bed... shouldn't be too hard to fix, though.
@1084747016 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The grammar for structs expects an identifier token for the member name, but the lexer is returning a typename token.
@1084747359 <andreas!andreas@cantor.gwydiondylan.org> Ok, time for bed. Night!
@1084747756 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Anybody here familiar with Melange's --framework option?
@1084747779 <bhoult!~bhoult@morannon.eservglobal.co.nz> no.  I think Rob added it.
@1084747887 -zelazny.freenode.net- hannes_ (~hannes@pD9E79F8E.dip.t-dialin.net) joined the channel
@1084747900 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Okay, I'll ask him. The thing is, it's tedious to have to supply the name of every framework that's used directly or indirectly, and it isn't how gcc works.
@1084747944 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought I'd align it with gcc (and CodeWarrior) behavior, but I wanted to check whether there's a reason for the current behavior.
@1084748423 <prom!~prom@217.13.206.33> re
@1084749882 <bhoult!~bhoult@morannon.eservglobal.co.nz> cpage: you haven't been sending p p p Powerbooks to London, have you?
@1084749932 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION looking at your iChat thumbnail
@1084750224 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> heh
@1084756113 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084756189 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> hi.  anyone else having trouble getting to the gd web pages?
@1084756205 <housel!housel@cantor.gwydiondylan.org> yeah, the DNS is down
@1084756226 <housel!housel@cantor.gwydiondylan.org> won't get rebooted until AM .de time
@1084756236 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> oh, ok
@1084756249 <bhoult!~bhoult@morannon.eservglobal.co.nz> meantime you can use 212.40.165.86
@1084756262 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> ah, thanks!
@1084756342 <bhoult!~bhoult@morannon.eservglobal.co.nz> if you put that in /etc/hosts then cvs etc will work too
@1084756378 <housel!housel@cantor.gwydiondylan.org> unless you're in windows, in which case you have to remember where the hosts file is
@1084756401 <bhoult!~bhoult@morannon.eservglobal.co.nz> doesn't it vary from version to version?
@1084756415 <housel!housel@cantor.gwydiondylan.org> maybe
@1084756455 <housel!housel@cantor.gwydiondylan.org> yes, it does:
@1084756456 <prom!~prom@217.13.206.33> *sigh* i see what the dns problem comes from. andreas still didnt replace the primary DNS, which is in a dead subnet.
@1084756457 <housel!housel@cantor.gwydiondylan.org> Windows 95/98/Me c:\windows\hosts
@1084756457 <housel!housel@cantor.gwydiondylan.org> Windows NT/2000/XP Pro  c:\winnt\system32\drivers\etc\hosts
@1084756457 <housel!housel@cantor.gwydiondylan.org> Windows XP Home c:\windows\system32\drivers\etc\hosts
@1084757434 <bhoult!~bhoult@morannon.eservglobal.co.nz> wow, stranhe that it varies between xp pro and xp home
@1084757452 <bhoult!~bhoult@morannon.eservglobal.co.nz> toplevel named as in 95/98, but contents are as in NT-series?
@1084758066 <housel!housel@cantor.gwydiondylan.org> I don't buy that actually, I'm in XP pro right now and it uses C:\WINDOWS
@1084758153 <bhoult!~bhoult@morannon.eservglobal.co.nz> well I sure don't have a clue -- I've only used 3 of those 7 OS's
@1084758268 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> Windows XP Pro is apparently the same as XP Home
@1084758572 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> oh, i didn't see you said they were different.  anyway, mine is XP Pro and it's in \Windows\System32\drivers\etc
@1084762132 <housel!housel@cantor.gwydiondylan.org> I've used all 7
@1084762182 <housel!housel@cantor.gwydiondylan.org> (plus 3.0, 3.1, and WFW 3.11)
@1084762480 <bhoult!~bhoult@morannon.eservglobal.co.nz> well I've used 4 if you count both NT3.5.1 and NT4 (which the original list doesn't seem to)
@1084762492 <bhoult!~bhoult@morannon.eservglobal.co.nz> plus 95 and 2K
@1084766468 <bhoult!~bhoult@morannon.eservglobal.co.nz> what is the purpose of /etc/cron.weekly/find ??
@1084766490 <bhoult!~bhoult@morannon.eservglobal.co.nz> it's using all the cpu time, and is up to 482 minutes of CPU time so far
@1084766510 <bhoult!~bhoult@morannon.eservglobal.co.nz> pid 22406
@1084767373 <bhoult!~bhoult@morannon.eservglobal.co.nz> ah, I see.  It's updating the "locate" database.  looks stuck, though.
@1084767526 <housel!housel@cantor.gwydiondylan.org> how could it possibly get stuck?
@1084768662 <bhoult!~bhoult@morannon.eservglobal.co.nz> I don't know
@1084768684 <bhoult!~bhoult@morannon.eservglobal.co.nz> could it be following a symlink to a point higher int he dir tree?
@1084768736 <bhoult!~bhoult@morannon.eservglobal.co.nz> could 8.6 hours (and counting) of CPU time for a "find" possibly be correct?
@1084769649 -zelazny.freenode.net- hannes_ (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084770203 <housel!housel@cantor.gwydiondylan.org> One wouldn't think so, then again there could be some seriously nested directory structure somewhere
@1084770491 <housel!housel@cantor.gwydiondylan.org> the current locatedb is 12076451 bytes
@1084770822 <bhoult!~bhoult@morannon.eservglobal.co.nz> I'm also confused about why there is an lpd that has used 964 min of CPU time
@1084770965 <housel!housel@cantor.gwydiondylan.org> ah yes that... somebody (brent) accidentally tried to print something a few months ago, and so it's still trying to send those jobs to a nonexistent printer
@1084771084 <prom!~prom@217.13.206.33> user brent has 22 open shells !?
@1084771095 <prom!~prom@217.13.206.33> lshd doesnt write utmp.... *sigh*
@1084771140 <housel!housel@cantor.gwydiondylan.org> I removed the print jobs and disabled the print queue
@1084771164 <prom!~prom@217.13.206.33> this machine has to be re-setup. andreas agreed. ill poke him to let me get started with it...
@1084771197 <prom!~prom@217.13.206.33> housel: i think you could just purge lpd... there is no printer there. this must have been left over from the base install.
@1084771647 <prom!~prom@217.13.206.33> w00000t. this box has all kinds of ports open. this machine would make a good administrative horror movie.
@1084771676 <housel!housel@cantor.gwydiondylan.org> heh
@1084771875 <prom!~prom@217.13.206.33> also, i think pserver is a bad idea. anonymous ssh is the way to go. savannah switched a while ago. but i think stuff like that can be done after the resetup.
@1084771926 <housel!housel@cantor.gwydiondylan.org> yeah I guess so
@1084772121 <bhoult!~bhoult@morannon.eservglobal.co.nz> well the find is up to 9.5 hours of CPU time now
@1084772136 <bhoult!~bhoult@morannon.eservglobal.co.nz> does anyone actually use locate?
@1084772159 <housel!housel@cantor.gwydiondylan.org> I use it, but it's not worth *that* high a price
@1084772177 <prom!~prom@217.13.206.33> bhoult: for me, the question is who got a stupid umask ;)
@1084772307 [prom!~prom@217.13.206.33] ACTION uses locate on all kinds of occasions. there is no real reason for not having it. exposing big build trees to it is just not necessary. what else could there be that has so many files?
@1084773871 <housel!housel@cantor.gwydiondylan.org> hmmm... I think the MPS disassembly code can't properly handle "mov edi,dword ptr [ebx+edi+0Bh]"
@1084773929 [prom!~prom@217.13.206.33] ACTION goes home. bbl.
@1084774946 -dircproxy- You disconnected
@1084775190 -dircproxy- You connected
@1084777315 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084777335 <housel!housel@cantor.gwydiondylan.org> morning
@1084777535 <andreas!andreas@cantor.gwydiondylan.org> dylan_wrapper_check is called from everywhere. Just disabling the test in onw or two places doesn't help.
@1084777555 <andreas!andreas@cantor.gwydiondylan.org> I really wonder why the compiler itself doesn't trigger the bug, but koala does.
@1084777563 <housel!housel@cantor.gwydiondylan.org> that must mean the wrapper really is invalid in this case
@1084777703 <housel!housel@cantor.gwydiondylan.org> I think fmtdy.c describes what dylan wrappers should look like
@1084777722 <housel!housel@cantor.gwydiondylan.org> bedtime for me... good night
@1084778239 <andreas!andreas@cantor.gwydiondylan.org> Night Peter!
@1084784222 -zelazny.freenode.net- brucehoult (~bruce@202.0.59.204) joined the channel
@1084785634 -zelazny.freenode.net- brucehoult_ (~bruce@202.0.59.121) joined the channel
@1084797132 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084797155 <brucehoult!~bruce@202.0.59.121> morning!
@1084797161 <andreas!andreas@cantor.gwydiondylan.org> Morning Bruce!
@1084797186 <brucehoult!~bruce@202.0.59.121> we going to get that DNS kicked in the guts soon?
@1084797225 <andreas!andreas@cantor.gwydiondylan.org> Didn't reach anybody yet. There were rumors about the ISP being in financial trouble, but I had expected to get advance warning...
@1084797249 <andreas!andreas@cantor.gwydiondylan.org> This is my personal mail server too, so the issue is sufficiently urgent for me too.
@1084797297 <brucehoult!~bruce@202.0.59.121> well cantor itself is perfectly reachable and working fine
@1084797330 <brucehoult!~bruce@202.0.59.121> by IP
@1084797332 <andreas!andreas@cantor.gwydiondylan.org> Yes, it is, and of course we could run the DNS there. It's just that redirecting DNS entries is difficult without being able to receive mail.
@1084797340 <andreas!andreas@cantor.gwydiondylan.org> Rats.
@1084797378 <brucehoult!~bruce@202.0.59.121> ah.  your incoming mail won't get there.  that *is* a problem
@1084797743 -zelazny.freenode.net- pritlove (~pritlove@pD9E6B641.dip.t-dialin.net) joined the channel
@1084799812 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning
@1084806048 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084807794 <cgay!~cgay@ita4fw1.itasoftware.com> g'day
@1084808621 <housel!housel@cantor.gwydiondylan.org> morning carl
@1084808653 <cgay!~cgay@ita4fw1.itasoftware.com> hey.  how goes the MPS battle?
@1084808727 <housel!housel@cantor.gwydiondylan.org> it's mostly working, but it looks like there are cases that the page fault handler doesn't handle properly
@1084808737 <housel!housel@cantor.gwydiondylan.org> Shouldn't be too hard to fix
@1084808805 <cgay!~cgay@ita4fw1.itasoftware.com> cool.  did you have to make changes in both the fd runtime and MPS?
@1084808813 <housel!housel@cantor.gwydiondylan.org> I got a letter from nick barnes, saying that the generation configuration is reasonable, and that they don't have a tarball release planned
@1084808832 <housel!housel@cantor.gwydiondylan.org> So far all the changes were in collector.c
@1084808843 <housel!housel@cantor.gwydiondylan.org> Though this bug I'm chasing might be in MPS
@1084808881 <housel!housel@cantor.gwydiondylan.org> (not a bug really, but maybe an unhandled case)
@1084809007 <cgay!~cgay@ita4fw1.itasoftware.com> i c.   well i was just thinking it would be simpler if they're in our runtime so we don't have to get changes back into the MPS source
@1084810214 -dircproxy- You disconnected
@1084811397 -dircproxy- You connected
@1084811713 <cgay!~cgay@ita4fw1.itasoftware.com> i'm thinking of adding a generic funcion "file-system-case-sensitive?" to the file-system module, but i'm not 100% sure it's justified yet.  i.e., does the fact that i find it necessary mean there's a limitation to the locators library that should be fixed...
@1084811852 [cgay!~cgay@ita4fw1.itasoftware.com] ACTION goes to check the URL RFC to verify that the 'path' portion of a URL can be case-(in)sensitive depending on platform...
@1084812572 <cgay!~cgay@ita4fw1.itasoftware.com> ah, i guess they are case-sensitive always.  thought i remembered it depending on OS.
@1084812715 <housel!housel@cantor.gwydiondylan.org> a client could never tell whether the server would be case-sensitive or not
@1084812774 <cgay!~cgay@ita4fw1.itasoftware.com> i'm just worried about doing regex matching on the URLs in the correct way
@1084813152 <cgay!~cgay@ita4fw1.itasoftware.com> also, whether on windows a filename that matches (but with the wrong case) should be 404...
@1084813392 <housel!housel@cantor.gwydiondylan.org> in practice, I imagine it usually isn't a 404 if it can actually find the file, but if matches in config files are involved then all bets are off
@1084813397 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084818181 <cgay!~cgay@ita4fw1.itasoftware.com> oh great.  now i'm getting versioncache problems here too
@1084818194 <cgay!~cgay@ita4fw1.itasoftware.com> they're contagious!
@1084818225 <housel!housel@cantor.gwydiondylan.org> don't worry about the messages, I don't think anything is affected
@1084818241 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1084818294 <cgay!~cgay@ita4fw1.itasoftware.com> my problem is slightly different, but seems related to the versioncache generation
@1084818312 <cgay!~cgay@ita4fw1.itasoftware.com> losing group write perms on the data dir
@1084818350 <housel!housel@cantor.gwydiondylan.org> it's not just group write perm, it's also other read perms
@1084818716 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084818770 <andreas!andreas@cantor.gwydiondylan.org> I have made substantial progress with the MPS problem that is triggered by koala.
@1084818792 <housel!housel@cantor.gwydiondylan.org> pray tell!
@1084818804 <andreas!andreas@cantor.gwydiondylan.org> The very first call to primitive-allocate-wrapper fails.
@1084818836 <andreas!andreas@cantor.gwydiondylan.org> The wrapper passed is <mm-wrapper>.class-instance-header, which turns out to be $undefined.
@1084818865 <andreas!andreas@cantor.gwydiondylan.org> So it is a problem of order of startup initializations, probably triggered by some construct in koala.
@1084818888 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084818910 <andreas!andreas@cantor.gwydiondylan.org> I suspect it's this line:
@1084818912 <andreas!andreas@cantor.gwydiondylan.org> define constant log-copious = curry(log-message, $log-copious);
@1084818946 <housel!housel@cantor.gwydiondylan.org> if you change that to an explicit function definition does it work?
@1084818951 <andreas!andreas@cantor.gwydiondylan.org> But that's only speculation based on seeing this symbol at one point in the compiler.
@1084818961 <andreas!andreas@cantor.gwydiondylan.org> Didn't try yet, will do next.
@1084818962 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084818965 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084818974 <andreas!andreas@cantor.gwydiondylan.org> Hi Rob!
@1084819050 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Andreas. 
@1084819063 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> l4 comes with a ppc emulator you can run it on. :-)
@1084819107 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Don't think it'll do HURD yet, but it's something to play with anyway.
@1084819118 <andreas!andreas@cantor.gwydiondylan.org> I happen to be able to use the vmware license my employer bought. Befre, I was using bochs for experiments.
@1084819150 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I just need a PPC emulator. Well, want. :-)
@1084819171 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I use VPC extensively at work to switch between Linux and Windows distributions.
@1084819475 <andreas!andreas@cantor.gwydiondylan.org> Carl: Are you still here?
@1084819482 <cgay!~cgay@ita4fw1.itasoftware.com> yes, hi
@1084819526 <andreas!andreas@cantor.gwydiondylan.org> Carl: I think you need a generic mechanism to make sure an attacker cannot trick his way through protection mechanisms by using different names for the same file.
@1084819538 <andreas!andreas@cantor.gwydiondylan.org> Are you aware of the poisoned NULL byte?
@1084819554 <cgay!~cgay@ita4fw1.itasoftware.com> no, but it sounds bad :)
@1084819649 <andreas!andreas@cantor.gwydiondylan.org> Imagine a web server written in a language that keeps expÃlicit track of string lengths. Like Dylan. Running on top of a OS with a system interface inspired by C, such as UNIX oder Windows, where strings are NULL-terminated.
@1084819669 <andreas!andreas@cantor.gwydiondylan.org> Now imagine what happens when you handle a request like foo.exe%00.txt
@1084819746 <cgay!~cgay@ita4fw1.itasoftware.com> you pass that string to the OS open() and all hell breaks loose?
@1084819791 <andreas!andreas@cantor.gwydiondylan.org> What usually happens is that you map from filename extensions to MIME types, and from there to request handlers.
@1084819814 <andreas!andreas@cantor.gwydiondylan.org> Say that .exe maps to application/cgi-bin, and .txt maps to text/plain.
@1084819833 <andreas!andreas@cantor.gwydiondylan.org> YOu have that CGI there, which of course contains your database passwords.
@1084819887 <andreas!andreas@cantor.gwydiondylan.org> The %00 gets converted into a NULL byte, which is a legal part of the string. SO your code thinks the user requests a txt file, and calls the appropriate handler.
@1084819905 <andreas!andreas@cantor.gwydiondylan.org> Which calls open() on the string it gets, and delivers the contents to the user.
@1084819923 <andreas!andreas@cantor.gwydiondylan.org> Of course, at this moment, the .txt is lost, and your server happily delivers the application binary.
@1084819923 <^self!~fn@66.209.66.105> i see.
@1084819938 <cgay!~cgay@ita4fw1.itasoftware.com> ah
@1084819996 <andreas!andreas@cantor.gwydiondylan.org> WIndows has a couple more gotchas like that. There once was the problem that you could append $00 to a filename to do some magic, and that $00 was stripped by the OS before opening the file.
@1084820021 <andreas!andreas@cantor.gwydiondylan.org> So you need a generic pathname canonicalization mechanism to protect against this problem in a systematic manner.
@1084820046 <andreas!andreas@cantor.gwydiondylan.org> This needs to be carefully worked out for every platform, including all of these gotchas.
@1084820079 -dircproxy- You disconnected
@1084820090 <cgay!~cgay@ita4fw1.itasoftware.com> i c.  do you have a good source for this info?  otherwise i'll just search for 'poisoned null byte'
@1084820133 <cgay!~cgay@ita4fw1.itasoftware.com> i wonder if CL-HTTP has already done something like this...
@1084820158 <andreas!andreas@cantor.gwydiondylan.org> I have explained all there is to know about that. The hard part is coming up with a design that allows for central maintenance of platform rules, so the user code doesn't have to worry anymore.
@1084820175 <andreas!andreas@cantor.gwydiondylan.org> You need to fight the whole class of bugs, not just this single instance.
@1084820303 <cgay!~cgay@ita4fw1.itasoftware.com> yup
@1084820389 <andreas!andreas@cantor.gwydiondylan.org> I've seen that you did well on preventing directory traversal, btw. :)
@1084820445 <andreas!andreas@cantor.gwydiondylan.org> There are web servers you can fool with %2e%2f%2f
@1084820458 <andreas!andreas@cantor.gwydiondylan.org> Err... %2e%2e%2f that is.
@1084820481 <cgay!~cgay@ita4fw1.itasoftware.com> which must be ../
@1084820530 <cgay!~cgay@ita4fw1.itasoftware.com> I'll make a note of your comments and try to get to that soon
@1084820585 <andreas!andreas@cantor.gwydiondylan.org> Cool. :) Meanwhile, I'll continue to tame MPS for koala on Linux. It's progressing...
@1084820670 <cgay!~cgay@ita4fw1.itasoftware.com> I almost have virtual hosts working, and a sort of generic directory policy mechanism
@1084820698 <andreas!andreas@cantor.gwydiondylan.org> I mention unification of legacy composite characters in UNICODE as another joyful problem in the same category.
@1084820746 <andreas!andreas@cantor.gwydiondylan.org> Just for your notes, not that we have much of UNICODE support at the moment. :)
@1084821044 <andreas!andreas@cantor.gwydiondylan.org> Regarding the directory policy: do you have a notion of a generic protection policy?
@1084821045 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> URL-encoding characters used to be good for getting the source to ASP pages. It was particularly bad as they'd already "ficed" the bug by refusing the ASCII that caused the same problem. :-)
@1084821052 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> "fixed", even.
@1084821393 <andreas!andreas@cantor.gwydiondylan.org> Oh, koala comes with yet another implementation of string-to-float! :)
@1084821414 <cgay!~cgay@ita4fw1.itasoftware.com> i think i stole it from chris double
@1084821471 <cgay!~cgay@ita4fw1.itasoftware.com> haven't done much of anything with security yet
@1084821540 <andreas!andreas@cantor.gwydiondylan.org> There's string-to-float and float-to-string in common-dylan. The Gwydion implementation is carefully written to be precise, in the sense that concatenate(string-to-float, float-to-string) and vice versa is the identity function.
@1084821616 <cgay!~cgay@ita4fw1.itasoftware.com> ok.  i also need to stop using the copies of anaphora, meta, xml-parser, etc that are in the koala dir
@1084821653 <andreas!andreas@cantor.gwydiondylan.org> Probably. Which reminds me taht we need some generic library distribution and installation mechanism.
@1084821676 <andreas!andreas@cantor.gwydiondylan.org> Sooner or later. :)
@1084821736 <cgay!~cgay@ita4fw1.itasoftware.com> do you mean something like CPAN?
@1084821745 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084821768 <andreas!andreas@cantor.gwydiondylan.org> CPAN was a major factor in perl's success.
@1084821790 <cgay!~cgay@ita4fw1.itasoftware.com> i've used it a lot lately.  it's probably the best thing about perl.  ;-)
@1084821828 <andreas!andreas@cantor.gwydiondylan.org> Oh, it has powerful code obfuscation facilities for job security. :)
@1084821891 <cgay!~cgay@ita4fw1.itasoftware.com> at this point i'd settle for a manually maintained library repository, which could be automated later
@1084821900 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084821913 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084822029 <cgay!~cgay@ita4fw1.itasoftware.com> does CPAN do dependency tracking?
@1084822038 <andreas!andreas@cantor.gwydiondylan.org> Yes.
@1084822089 <cgay!~cgay@ita4fw1.itasoftware.com> so it wouldn't install a too-new version...
@1084822128 <andreas!andreas@cantor.gwydiondylan.org> Not sure if they can do that.
@1084822182 <andreas!andreas@cantor.gwydiondylan.org> Something like a too-new version shouldn't happen. If a version breaks backwards compatibility, that is a major version change, which warrants a new, non-conflicting library name, so you can have both versions coexist for old and new code.
@1084822394 -zelazny.freenode.net- hannes_ (~hannes@pD9E79F8E.dip.t-dialin.net) joined the channel
@1084822395 <hannes_!~hannes@pD9E79F8E.dip.t-dialin.net> re
@1084822415 <andreas!andreas@cantor.gwydiondylan.org> Re Hannes.
@1084822422 <hannes_!~hannes@pD9E79F8E.dip.t-dialin.net> hi andreas
@1084822472 -dircproxy- You connected
@1084822472 <andreas!andreas@cantor.gwydiondylan.org> Ok, the error happens when define-class(<singleton-class>) is run an program startup time, while finalizing the class inheritance.
@1084822854 <andreas!andreas@cantor.gwydiondylan.org> If I remove <singleton-class> from the code, the problem reappears during define-class for something in xml-rpc-common.
@1084823564 <andreas!andreas@cantor.gwydiondylan.org> In fact, the minimal code to trigger the bug is: *tada*
@1084823565 <andreas!andreas@cantor.gwydiondylan.org> define class <singleton-object> (<object>) end;
@1084823742 <andreas!andreas@cantor.gwydiondylan.org> The thing is: the whole compiler is built with itself, and works fine. The only difference is that the compiler is built as a system project, where the above is compiled in a user project.
@1084823792 -dircproxy- You disconnected
@1084824176 <prom!~prom@217.13.206.33> re
@1084824192 <andreas!andreas@cantor.gwydiondylan.org> re prom.
@1084825098 -dircproxy- You connected
@1084825315 -zelazny.freenode.net- jdrake (irc_user@CPE00045afdd0e8-CM014410113717.cpe.net.cable.rogers.com) joined the channel
@1084825329 <housel!housel@cantor.gwydiondylan.org> are you sure? the current build instructions in koala.html have it being built as a system project using the registry
@1084825336 <jdrake!irc_user@CPE00045afdd0e8-CM014410113717.cpe.net.cable.rogers.com> anything new?
@1084825350 <housel!housel@cantor.gwydiondylan.org> since when?
@1084825368 <andreas!andreas@cantor.gwydiondylan.org> Peter: I don't follow those. Maybe I should. :)
@1084825371 <jdrake!irc_user@CPE00045afdd0e8-CM014410113717.cpe.net.cable.rogers.com> Since thy holy god created Dylan...
@1084825374 <andreas!andreas@cantor.gwydiondylan.org> Hi jdrake.
@1084825397 <andreas!andreas@cantor.gwydiondylan.org> About a million lines of source code in our CVS, I guess.
@1084825521 <andreas!andreas@cantor.gwydiondylan.org> Hm. The problem doesn't go away when compiled as a system library.
@1084825539 <housel!housel@cantor.gwydiondylan.org> so that's not it then
@1084825581 <housel!housel@cantor.gwydiondylan.org> how about using LD_PRELOAD to insure that libdylan.so gets loaded first?
@1084825626 <housel!housel@cantor.gwydiondylan.org> what does ldd say?
@1084825637 <andreas!andreas@cantor.gwydiondylan.org> Might bw worth a try. But I still want to figure out what the difference between my one line of code and all the rest of the whole compiler is that makes the one line crash.
@1084825643 <andreas!andreas@cantor.gwydiondylan.org> ldd looks good.
@1084825789 <andreas!andreas@cantor.gwydiondylan.org> I'm very relieved that this looks like it is *not* a gc-related problem.
@1084825901 <housel!housel@cantor.gwydiondylan.org> my "jam-build" program works fine when compiled with MPS
@1084826038 <andreas!andreas@cantor.gwydiondylan.org> Ok, it is one of the lines in the .hdp file that triggers the problem.
@1084826111 <housel!housel@cantor.gwydiondylan.org> what line?
@1084826128 <andreas!andreas@cantor.gwydiondylan.org> I'm trying to find out right that.
@1084827233 <andreas!andreas@cantor.gwydiondylan.org> I was distracted by a new bug. When compiling a project from the command line of minimal-console-compiler, I get:
@1084827234 <andreas!andreas@cantor.gwydiondylan.org> Internal error: : 4 is not of type {<class>: <class>}
@1084827248 <andreas!andreas@cantor.gwydiondylan.org> during linking. Batch compiling works fine, though.
@1084827770 <andreas!andreas@cantor.gwydiondylan.org> How do I get a decent stack trace out of the compiler?
@1084828069 <housel!housel@cantor.gwydiondylan.org> I'm not sure
@1084828114 <housel!housel@cantor.gwydiondylan.org> I'd find where the Internal error: is being printed, and make it call break()
@1084831427 <andreas!andreas@cantor.gwydiondylan.org> Hm. I was able to get rid of the bug in two libraries by changing compiler mode from loose to tight.
@1084831544 <housel!housel@cantor.gwydiondylan.org> hmmm indeed
@1084831557 <andreas!andreas@cantor.gwydiondylan.org> Yes! Removing a call to limited in koala finally gives some startup!
@1084831672 <cgay!~cgay@ita4fw1.itasoftware.com> interesting.  i ran into some strange problems (method not congruent during add-method) when in tight mode
@1084831675 <andreas!andreas@cantor.gwydiondylan.org> But crashes as soon as I try to connect to the port.
@1084831771 <andreas!andreas@cantor.gwydiondylan.org> Ok, but we're getting there!
@1084832556 -zelazny.freenode.net- hannes (~hannes@pD9E79F8E.dip.t-dialin.net) joined the channel
@1084833074 <andreas!andreas@cantor.gwydiondylan.org> Ok, the bug now is that the network library doesn't properly handle EINTR.
@1084833208 <housel!housel@cantor.gwydiondylan.org> why does the network library have to handle EINTR?
@1084833233 <andreas!andreas@cantor.gwydiondylan.org> Because the garbage collector throws around signals, interrupting the syscall to accept().
@1084833251 <housel!housel@cantor.gwydiondylan.org> ah yes
@1084833273 <housel!housel@cantor.gwydiondylan.org> should be easy to fix though
@1084833279 <prom!~prom@217.13.206.33> yesyes. tasty unixy syscall interruption.
@1084833329 <housel!housel@cantor.gwydiondylan.org> "define interruptible-syscall accept ... end"
@1084833419 <housel!housel@cantor.gwydiondylan.org> unix-socket-accessor.dylan is full of comments describing windows return codes
@1084833545 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) left the channel
@1084834194 -zelazny.freenode.net- hannes__ (~hannes@pD9E7A4C8.dip.t-dialin.net) joined the channel
@1084834261 <andreas!andreas@cantor.gwydiondylan.org> One workaround later, and koala is happily serving pages to me.
@1084834480 <housel!housel@cantor.gwydiondylan.org> yaay!
@1084834569 <andreas!andreas@cantor.gwydiondylan.org> Remind you: I still see the crawling-to-a-halt problem with gctest, and I worked around the problem of initialization order for the wrapper wrapper.
@1084834588 <andreas!andreas@cantor.gwydiondylan.org> But I keep seeing milestones here. :)
@1084835137 <andreas!andreas@cantor.gwydiondylan.org> FINISHED --01:04:34--
@1084835137 <andreas!andreas@cantor.gwydiondylan.org> Downloaded: 231,622 bytes in 59 files
@1084835141 <andreas!andreas@cantor.gwydiondylan.org> Output from wget.
@1084835336 <housel!housel@cantor.gwydiondylan.org> it couldn't do that with Boehm?
@1084835379 <andreas!andreas@cantor.gwydiondylan.org> Nope. The startup race condition I've been hunting for weeks!
@1084835952 <andreas!andreas@cantor.gwydiondylan.org> Ok, when I run wget in an endless loop, koala eventually crashes. After serving thousands of requests, though.
@1084836144 <housel!housel@cantor.gwydiondylan.org> could that be due to some other syscall being interrupted?
@1084836199 <andreas!andreas@cantor.gwydiondylan.org> No. The crash destroys the pthread structures. The stack looks smashed too.
@1084836226 <andreas!andreas@cantor.gwydiondylan.org> Again, the crashes are threads-related. The same I saw with the other garbage collector.
@1084836293 <andreas!andreas@cantor.gwydiondylan.org> I have an uneasy feeling about the runtime blatantly placing the current_teb pointer at a fixed address from the stack bottom. I mean, this memory range has been allocated for some purpose, probably for pthread internal data structures!
@1084836359 <andreas!andreas@cantor.gwydiondylan.org> Anyways, I think I've made progress, and need to go to bed. Night!
@1084836382 <housel!housel@cantor.gwydiondylan.org> 'night
@1084836811 <bhoult!~bhoult@morannon.eservglobal.co.nz> our friendly "find" is up to 27 hours CPU
@1084836862 <housel!housel@cantor.gwydiondylan.org> gack
@1084836874 <bhoult!~bhoult@morannon.eservglobal.co.nz> enough is enough.  it's dying.  now.
@1084836930 <bhoult!~bhoult@morannon.eservglobal.co.nz> wonder how much output it produced.  I'd better keep an eye on that "sort"
@1084836979 <bhoult!~bhoult@morannon.eservglobal.co.nz> noope .. it finished quickly, as did mandb
@1084837017 <bhoult!~bhoult@morannon.eservglobal.co.nz> still no DNS?
@1084837041 <housel!housel@cantor.gwydiondylan.org> no
@1084837088 -zelazny.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) joined the channel
@1084840127 -zelazny.freenode.net- cpage_ (~Chris@208.240.243.170) joined the channel
@1084840445 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084840793 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084841294 -zelazny.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1084842438 -dircproxy- You disconnected
@1084843896 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1084843939 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> is gwydiondylan.org down?
@1084844207 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> DNS isn't working. You can reach the server by IP: 212.40.165.86
@1084844359 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> I see
@1084844579 <bhoult!~bhoult@morannon.eservglobal.co.nz> we're working on restoring full service... but I don't think we have an ETA
@1084844741 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> hosting problem?
@1084845350 -zelazny.freenode.net- jdrake (irc_user@CPE00045afdd0e8-CM014410113717.cpe.net.cable.rogers.com) joined the channel
@1084846100 -dircproxy- You connected
@1084846773 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) left the channel
@1084846848 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1084848154 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084849810 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) left the channel
@1084853370 -dircproxy- You disconnected
@1084853384 -dircproxy- You connected
@1084855941 -zelazny.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1084864563 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning *
@1084864856 -zelazny.freenode.net- brucehoult (~bruce@202-0-61-22.adsl.paradise.net.nz) joined the channel
@1084866162 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1084866173 <brucehoult!~bruce@202-0-61-22.adsl.paradise.net.nz> re Chris
@1084866972 -dircproxy- You disconnected
@1084867789 <gabor!root@cantor.gwydiondylan.org> re.
@1084870366 -zelazny.freenode.net- brucehoult (~bruce@202-0-61-78.adsl.paradise.net.nz) joined the channel
@1084873656 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084873683 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> morning
@1084873832 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> any news on DNS?
@1084873852 <andreas!andreas@cantor.gwydiondylan.org> Nope. WIll start my calls right now. :/
@1084874087 <andreas!andreas@cantor.gwydiondylan.org> Finally, a techie at the ISP is on the phone! Now let's hope the machine isn't physically broken...
@1084874698 <andreas!andreas@cantor.gwydiondylan.org> Ok, the machine doesn't make a broken impression: fans are ok, hard disks too. It's probably right in the middle of an fsck now. If it isn't up within a quarter of an hour, I'll call again.
@1084875048 <andreas!andreas@cantor.gwydiondylan.org> Ok, the host is back up. The vmwares need manual intervention, IIRC...
@1084875098 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> what machine is this?
@1084875137 <andreas!andreas@cantor.gwydiondylan.org> adrenochrome. A system I'm operationg in Amsterdam for a couple of friends. It is also my backup DNS and mail server.
@1084875159 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> ah
@1084875188 <andreas!andreas@cantor.gwydiondylan.org> My main mail and DNS server is unreachable, since the ISP went bankrupt.
@1084875205 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> :-(
@1084875653 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1084875660 <andreas!andreas@cantor.gwydiondylan.org> Hi hannes.
@1084875679 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi andreas
@1084876592 <andreas!andreas@cantor.gwydiondylan.org> Ok, DNS is back up.
@1084876604 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> oh
@1084876605 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> I just got a first reply from "host www.gwydiondylan.org"
@1084876634 <brucehoult!~bruce@202-0-61-78.adsl.paradise.net.nz> 4 times it timed out, I thougth the 5th was going to too, but eventually worked 
@1084876654 <andreas!andreas@cantor.gwydiondylan.org> Yes, it took a while to fsck all the file systems.
@1084884165 -zelazny.freenode.net- cpage (~Chris@67.118.119.204) joined the channel
@1084887413 <cgay!~cgay@ita4fw1.itasoftware.com> looks like andreas made good progress yesterday.  woo hoo
@1084887484 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> cgay: yes, koala was much faster and much more stable with mps :) (i saw it on andreas laptop)
@1084887759 <cgay!~cgay@ita4fw1.itasoftware.com> great!
@1084889606 <andreas!andreas@cantor.gwydiondylan.org> Hi Carl, re Hannes!
@1084889710 <andreas!andreas@cantor.gwydiondylan.org> What really needs work now is the error handling of all the syscalls in the network library, as well as proper error handling in koala itself.
@1084890028 -dircproxy- You connected
@1084890532 <housel!housel@cantor.gwydiondylan.org> Nick Barnes gave me the Dylan MM requirements document; it's up at http://www.gwydiondylan.org/~housel/req.dylan.txt
@1084890578 <andreas!andreas@cantor.gwydiondylan.org> Cool!
@1084890907 <andreas!andreas@cantor.gwydiondylan.org> Hm. The 1270th thread koala is spawning never makes it through startup, hanging in thre pthread start, waiting for a signal that never arrives.
@1084890946 <andreas!andreas@cantor.gwydiondylan.org> I have reproduced that exact count numerous times, with either one or two wget processed generating load.
@1084890971 <housel!housel@cantor.gwydiondylan.org> that's a strange number
@1084891000 <andreas!andreas@cantor.gwydiondylan.org> Yes indeed.
@1084891572 <andreas!andreas@cantor.gwydiondylan.org> Ah, I think I know where the problem comes from. ps shows a lot of koala zombie processes. So the threads are never correctly terminated.
@1084891652 <housel!housel@cantor.gwydiondylan.org> threads are terminated with join(), right? isn't there some sort of known limitation with join()?
@1084891682 <andreas!andreas@cantor.gwydiondylan.org> I think the creator has to call join. Threads terminate themselves by returning from the thread function.
@1084891737 <housel!housel@cantor.gwydiondylan.org> does the creator call join?
@1084891771 <housel!housel@cantor.gwydiondylan.org> (it's bug 6978 BTW)
@1084891897 <andreas!andreas@cantor.gwydiondylan.org> No, nobody is calling join.
@1084892115 <andreas!andreas@cantor.gwydiondylan.org> Interesting. No zombies when not running under gdb.
@1084892364 <housel!housel@cantor.gwydiondylan.org> does it die at 1270 when not under gdb?
@1084892460 <andreas!andreas@cantor.gwydiondylan.org> Hard to tell, because it was gdb keeping track of thread count. But PIDs should gve a clue...
@1084892655 <andreas!andreas@cantor.gwydiondylan.org> Ok, PID difference is > 1500, and it is still running.
@1084892665 <andreas!andreas@cantor.gwydiondylan.org> It did get slower, though.
@1084892779 <housel!housel@cantor.gwydiondylan.org> is it really a good idea to create that many threads?  Doesn't apache have a pool of live threads it passes things to?
@1084892797 <housel!housel@cantor.gwydiondylan.org> that would be less overhead, too
@1084892814 <housel!housel@cantor.gwydiondylan.org> (mutex wait overhead vs. thread startup overhead)
@1084892814 <andreas!andreas@cantor.gwydiondylan.org> Ok, and finally it hangs, with PID difference between first and last thread of about 1600.
@1084892888 <andreas!andreas@cantor.gwydiondylan.org> Still, you'd expect it to work. It's not that the 1000 threads are all active at the same time.
@1084892931 <housel!housel@cantor.gwydiondylan.org> yeah, but how are their stacks being allocated?
@1084892964 <housel!housel@cantor.gwydiondylan.org> does glibc notice they've gone away and re-use the stack space?
@1084892970 <andreas!andreas@cantor.gwydiondylan.org> Good question.
@1084892995 <andreas!andreas@cantor.gwydiondylan.org> It does.
@1084893077 <housel!housel@cantor.gwydiondylan.org> Does dylan_mm_deregister_thread_from_teb() get called for each thread that has gone away?
@1084893094 <andreas!andreas@cantor.gwydiondylan.org> I think so. Didn't verify it, though.
@1084893117 <andreas!andreas@cantor.gwydiondylan.org> The problem is in the communication between the newly created thread, and the pthread manager thread.
@1084893118 <^self!~fn@66.209.66.105> yowzalicious!
@1084893229 <andreas!andreas@cantor.gwydiondylan.org> Ok, have to leave. Later!
@1084893234 <housel!housel@cantor.gwydiondylan.org> cul
@1084895683 <CIA-2!micah@hyperreal.info> commit by andreas to src configure.in,1.152,1.153: bug: 7017
@1084895683 <CIA-2!micah@hyperreal.info> Better regexp for matching Linux on PC.
@1084895686 <CIA-2!micah@hyperreal.info> commit by andreas to src: bug: 7018
@1084895686 <CIA-2!micah@hyperreal.info> Add <value-cell> implementation.
@1084896073 -dircproxy- You disconnected
@1084899576 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084899590 <andreas!andreas@cantor.gwydiondylan.org> CIA obviously has lag.
@1084900313 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084900728 -dircproxy- You connected
@1084900806 <housel!housel@cantor.gwydiondylan.org> maybe it was refusing messages from gwydiondylan.org until DNS came up
@1084900913 <andreas!andreas@cantor.gwydiondylan.org> The commit was after DNS came up again.
@1084900934 <andreas!andreas@cantor.gwydiondylan.org> TODO: Real errno accessor
@1084900935 <andreas!andreas@cantor.gwydiondylan.org> define method unix-errno () 99 end;
@1084900978 <housel!housel@cantor.gwydiondylan.org> yes, I remember that
@1084900983 <andreas!andreas@cantor.gwydiondylan.org> I'm just working on implementing that. We need to handle EINTR, it's breaking everything, including stream reading and writing.
@1084901492 <gabor!root@cantor.gwydiondylan.org> use c-expr() ?
@1084901500 <gabor!root@cantor.gwydiondylan.org> on d2c
@1084901508 <andreas!andreas@cantor.gwydiondylan.org> Hi Gabor! No, FD in this case.
@1084901526 <gabor!root@cantor.gwydiondylan.org> ok
@1084901555 <housel!housel@cantor.gwydiondylan.org> fortunately for the network library you can use C-FFI
@1084901556 <gabor!root@cantor.gwydiondylan.org> then this sounds like a primitive
@1084901566 <gabor!root@cantor.gwydiondylan.org> ok
@1084901582 <andreas!andreas@cantor.gwydiondylan.org> Yes. Unfortunately, errno is implemented as a macro in glibc.
@1084901592 <housel!housel@cantor.gwydiondylan.org> no problem, write a C function to fetch it
@1084901602 <housel!housel@cantor.gwydiondylan.org> Use C-Source-File: errno.c in the LID file
@1084901610 <gabor!root@cantor.gwydiondylan.org> that is the portable way, yes
@1084901618 <andreas!andreas@cantor.gwydiondylan.org> I'll call __errno_location for now.
@1084901886 <housel!housel@cantor.gwydiondylan.org> streams and file-system will have to use primitive-* though, sigh
@1084901902 <housel!housel@cantor.gwydiondylan.org> though I think they already have code to retrieve errno properly
@1084901964 <andreas!andreas@cantor.gwydiondylan.org> Hm. I did see spurious read errors when two threads were active. I'll look into that later.
@1084902011 <housel!housel@cantor.gwydiondylan.org> I don't think they're chechking EINTR properly, but at least there is a way to retrieve errno
@1084902028 <andreas!andreas@cantor.gwydiondylan.org> Ok. So it won't be too hard to fix.
@1084905229 <cpage!~Chris@67.118.119.204> re
@1084905235 <andreas!andreas@cantor.gwydiondylan.org> Re Chris.
@1084905461 <andreas!andreas@cantor.gwydiondylan.org> Ok, handling EINTR everywhere solved the problem of koala becoming slower over time. Maybe there is a problem with non-local exits and signal delivery.
@1084905533 <andreas!andreas@cantor.gwydiondylan.org> It didn't solve the full-stop problem after 1270 threads. I think the problem is that nobody ever says join, and the thread isn't created detached either, so a descriptor is kept around.
@1084906316 <andreas!andreas@cantor.gwydiondylan.org> Ok, it seems that somebody needs to call pthread_detach if nobody is ever going to call pthread_join.
@1084906340 <andreas!andreas@cantor.gwydiondylan.org> Unfortunately, there's no notion of saying "I won't ever call join".
@1084906439 -zelazny.freenode.net- hannes (~hannes@pD9E7A4C8.dip.t-dialin.net) joined the channel
@1084906442 <hannes!~hannes@pD9E7A4C8.dip.t-dialin.net> re
@1084906627 <andreas!andreas@cantor.gwydiondylan.org> Re hannes.
@1084908391 -zelazny.freenode.net- hannes (~hannes@pD9E7A4C8.dip.t-dialin.net) joined the channel
@1084909617 <housel!housel@cantor.gwydiondylan.org> maybe a new init-keyword on <thread> would be appropriate
@1084909668 <andreas!andreas@cantor.gwydiondylan.org> I've settled for calling pthread_detach in the finalizer of <thread>, or actually in primitive_destroy_thread, which is called from there.
@1084909685 <andreas!andreas@cantor.gwydiondylan.org> If there's no reference left, nobody will be able to call join, right?
@1084909702 <andreas!andreas@cantor.gwydiondylan.org> Compile run to test the theory is on the way.
@1084912009 -zelazny.freenode.net- pritlove (~pritlove@217.231.164.200) joined the channel
@1084912241 -dircproxy- You disconnected
@1084913442 <andreas!andreas@cantor.gwydiondylan.org> Ok, looks like the finalizer isn't properly invoked. When calling it manually, koala becomes rock solid.
@1084913612 -dircproxy- You connected
@1084913634 <cgay!~cgay@ita4fw1.itasoftware.com> don't we have the same problem as java finalizers?  i.e., there's no guarantee when a gc will be done?
@1084913713 [cgay!~cgay@ita4fw1.itasoftware.com] ACTION should really just shut up since he doesn't have time to _think_
@1084913715 <andreas!andreas@cantor.gwydiondylan.org> That could very well be.
@1084913743 <andreas!andreas@cantor.gwydiondylan.org> BTW: I have koala running for 20 minutes straight now, serving about 50 requests per second.
@1084913784 <cgay!~cgay@ita4fw1.itasoftware.com> sah-weet :)
@1084913788 <housel!housel@cantor.gwydiondylan.org> How are you calling it manually?
@1084913796 <andreas!andreas@cantor.gwydiondylan.org> finalize(thread)
@1084913807 <housel!housel@cantor.gwydiondylan.org> ah
@1084914833 <housel!housel@cantor.gwydiondylan.org> I'll look into finalization this evening
@1084914864 <andreas!andreas@cantor.gwydiondylan.org> I have an instance here that looks like the finalizer was actually called.
@1084914943 <housel!housel@cantor.gwydiondylan.org> I imagine the threads are ending up in Generation 2, which is 2MB, and so it takes awhile for things to get cleaned out of that
@1084914995 <andreas!andreas@cantor.gwydiondylan.org> Might be. If you could figure out a way to fine-tune finalization behaviour, that would be good.
@1084915018 <andreas!andreas@cantor.gwydiondylan.org> Meanwhile, I just ignore the error from the second pthread_detach.
@1084916413 <CIA-2!micah@hyperreal.info> commit by andreas to fundev: Bug: 7019
@1084916413 <CIA-2!micah@hyperreal.info> Improve error handling somewhat. Especially EINTR is handled correctly
@1084916413 <CIA-2!micah@hyperreal.info> now (or so I hope).
@1084916805 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084916807 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084916817 <andreas!andreas@cantor.gwydiondylan.org> Hi Rob!
@1084916823 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION Software patents approved in EU http://kwiki.ffii.org/?Cons040518En :-<
@1084916830 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya Andreas.
@1084916860 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Was Troy any good? Work are dragging me to see Van Helsing next week. :-)
@1084916956 <andreas!andreas@cantor.gwydiondylan.org> The movie is about average, but you can safely take somebody there who enjoys a view of Brad Pitt or Orlando Bloom naked.
@1084917000 <cgay!~cgay@ita4fw1.itasoftware.com> heh
@1084917004 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I may need to restrain someone. :-)
@1084917246 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> What about that patent vote, though. What happened there? I thought Deutschland was going to do the sane thing. :-(
@1084917276 <andreas!andreas@cantor.gwydiondylan.org> I was thinking that too. The maneuver completely surprises me.
@1084917281 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION lives in the Fuedal part of the EU known as the UK.
@1084917324 <andreas!andreas@cantor.gwydiondylan.org> But what really boggles my mind is that the Commission is more powerful than the parliament. That's not how democracy is supposed to work.
@1084917332 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I think I'm going to patent patentability. Then any patentees will have to pay me to patent anything.
@1084917387 <housel!housel@cantor.gwydiondylan.org> time to really raise a protest
@1084917406 <housel!housel@cantor.gwydiondylan.org> for democratic governance, as well as patents
@1084918355 <bhoult!~bhoult@morannon.eservglobal.co.nz> "Better regexp for matching Linux on PC."
@1084918365 <bhoult!~bhoult@morannon.eservglobal.co.nz> Newsflash: that star was not a Kleene star...
@1084918378 <andreas!andreas@cantor.gwydiondylan.org> Eek!
@1084918964 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Eh?
@1084919020 <andreas!andreas@cantor.gwydiondylan.org> Globbing patterns are not regular expressions. I used the wrong terminology in my commit message.
@1084919032 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Ah.
@1084919061 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Regarding the Comission, yes the EU is run by unelected, unaccountable, corrupt bureacrats. Who have voted to hold themselves immune from corruption. And, joy, the UK sends their worst failed politicians to be the best of the corrupt (Kinnock, Patten).
@1084919103 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The EU is a wonderful ideal, but the Comission needs massive reform. Shame it would be in charge of any reform. :-)
@1084919215 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Manyhow, I'm sending some cash to FFII and writing to my MPs...
@1084919236 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION Bit late, obviously...
@1084919348 [robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com] ACTION rant, rant. Sorry.... :-)
@1084919373 <andreas!andreas@cantor.gwydiondylan.org> No problem. I share your concerns.
@1084919637 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Back on topic, would a flag to melange to make it declare functions rather than methods make sense? It might speed up the OpenGL bindings, for instance.
@1084919676 <housel!housel@cantor.gwydiondylan.org> I doubt that would make much of a difference
@1084919719 <housel!housel@cantor.gwydiondylan.org> the methods are going to be in sealed GFs
@1084919825 <bhoult!~bhoult@morannon.eservglobal.co.nz> +1 on ROb
@1084919850 <bhoult!~bhoult@morannon.eservglobal.co.nz> anything generated automatically from a C interface will *always* be able to be a simple function
@1084919873 <bhoult!~bhoult@morannon.eservglobal.co.nz> it will save a lot of codespace and link time to not have the general entries
@1084919882 <andreas!andreas@cantor.gwydiondylan.org> Except for struct accessors, if you drop the prefixes.
@1084919892 <housel!housel@cantor.gwydiondylan.org> exactly
@1084919916 <housel!housel@cantor.gwydiondylan.org> there are several gf struct accessors in the gtk1 binding
@1084919952 <andreas!andreas@cantor.gwydiondylan.org> There are zero struct accessors in the GTK2 binding. :)
@1084919989 <housel!housel@cantor.gwydiondylan.org> by design?
@1084919997 <bhoult!~bhoult@morannon.eservglobal.co.nz> It might also be good to have the option of the glue functions being inlined
@1084920020 <housel!housel@cantor.gwydiondylan.org> hopefully they won't be needed for DUIM
@1084920041 <housel!housel@cantor.gwydiondylan.org> the current gtk-duim did need them in several places
@1084920046 <andreas!andreas@cantor.gwydiondylan.org> Yes. The GTK2 API is fully usable without any access to structs.
@1084920072 <housel!housel@cantor.gwydiondylan.org> I suspected as much
@1084920075 <andreas!andreas@cantor.gwydiondylan.org> Accessing struct members directly is highly discouraged.
@1084920141 <housel!housel@cantor.gwydiondylan.org> that, and enum values, were the main reasons I had to write gobject-tool
@1084920551 -zelazny.freenode.net- pritlove (~pritlove@pD9E79E7A.dip.t-dialin.net) joined the channel
@1084920808 -zelazny.freenode.net- hannes_ (~hannes@pD9E79E7A.dip.t-dialin.net) joined the channel
@1084921118 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Inlined?
@1084921189 <housel!housel@cantor.gwydiondylan.org> that's a question?
@1084921214 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Yes. :-) What's inlined in a d2c context?
@1084921215 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> The options for --functions would be none , functions , all [functions & accessors]. So you could avoid accessors without prefixes, and manualy rename any collisions on functions.
@1084921387 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> housel: where does mps_lib_abort() in the runtime come from? it's used in collector.c, but i can't find a lib which provides it.
@1084921510 <housel!housel@cantor.gwydiondylan.org> in linux, it's provided by mpsplan.a
@1084921518 <housel!housel@cantor.gwydiondylan.org> in Win32, it's part of plinth.c
@1084921602 <housel!housel@cantor.gwydiondylan.org> (If memory serves me correctly)
@1084921674 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> yes, in win32/plinth.c i found it, but my mpsplan.a doesn't provide it and in my mps-sourcedir there is no match for mps_lib_abort.
@1084921820 <housel!housel@cantor.gwydiondylan.org> hmmm... I think I may have added one to one of the other files... let me check
@1084921901 <housel!housel@cantor.gwydiondylan.org> I added this to the beginning of linux-support.c:
@1084921903 <housel!housel@cantor.gwydiondylan.org> void mps_lib_abort(void)
@1084921903 <housel!housel@cantor.gwydiondylan.org> { 
@1084921903 <housel!housel@cantor.gwydiondylan.org>   fflush(stdout);
@1084921903 <housel!housel@cantor.gwydiondylan.org>   abort();
@1084921903 <housel!housel@cantor.gwydiondylan.org> }
@1084921993 <housel!housel@cantor.gwydiondylan.org> We should probably go the other direction and eliminate its use, it's not part of the mpslib.h interface anymore
@1084922019 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> thanks
@1084922922 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> how did you fix the linuxulator (to run mps on FreeBSD)?
@1084923067 <housel!housel@cantor.gwydiondylan.org> I wasn't able to fix that
@1084923092 <housel!housel@cantor.gwydiondylan.org> I hacked MPS to read the page fault address from the err field of sigcontext rather than the cr2 field
@1084923110 <housel!housel@cantor.gwydiondylan.org> I think there were 2 functions I had to change
@1084923130 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> housel: ah, ok. can you mail me a diff?
@1084923149 <housel!housel@cantor.gwydiondylan.org> ok
@1084923825 <housel!housel@cantor.gwydiondylan.org> hannes_: see http://www.gwydiondylan.org/~housel/mps-code.diff
@1084924079 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> you also use the version downloaded with wget from http://www.ravenbrook.com/project/mps/master/code/ ?
@1084924092 <housel!housel@cantor.gwydiondylan.org> yes
@1084924186 <hannes_!~hannes@pD9E79E7A.dip.t-dialin.net> andreas koala just responded to ~100 requests per second with a static file.
@1084924230 <housel!housel@cantor.gwydiondylan.org> nifty!
@1084925127 -dircproxy- You disconnected
@1084926333 -zelazny.freenode.net- hannes_ (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1084926866 -dircproxy- You connected
@1084927997 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1084928688 -dircproxy- You disconnected
@1084942108 -dircproxy- You connected
@1084942109 -dircproxy- You disconnected
@1084942242 -dircproxy- You connected
@1084943690 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084943691 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084944280 -zelazny.freenode.net- Enometh (~Dadaism@adsl-66-159-224-83.dslextreme.com) joined the channel
@1084944348 <housel!housel@cantor.gwydiondylan.org> morning rob
@1084944409 -zelazny.freenode.net- Enometh (~Dadaism@adsl-66-159-224-83.dslextreme.com) left the channel
@1084944923 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Morning. Early wake-up call from Liam this morning. :-)
@1084956183 -dircproxy- You disconnected
@1084956207 -dircproxy- You connected
@1084957221 -dircproxy- You disconnected
@1084957236 -dircproxy- You connected
@1084961293 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1084962086 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1084967862 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.200) joined the channel
@1084971764 -zelazny.freenode.net- pritlove (~pritlove@194.95.203.200) left the channel
@1084972160 <^self!~fn@66.209.66.105> hmm.  bored.
@1084973729 <gabor!root@cantor.gwydiondylan.org> go and read dilbert
@1084974043 <gabor!root@cantor.gwydiondylan.org> andreas: http://security.e-matters.de/advisories/072004.html
@1084975451 <^self!~fn@66.209.66.105> dilbert!  haven't read that in a while.  thanks.
@1084976140 <andreas!andreas@cantor.gwydiondylan.org> Re.
@1084976150 <andreas!andreas@cantor.gwydiondylan.org> koala on http://www.gwydiondylan.org:8000
@1084977406 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1084977427 <andreas!andreas@cantor.gwydiondylan.org> Hi Hannes.
@1084977494 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi andreas
@1084978194 <cgay!~cgay@ita4fw1.itasoftware.com> hi all
@1084978223 <andreas!andreas@cantor.gwydiondylan.org> Hi Carl!
@1084978244 <cgay!~cgay@ita4fw1.itasoftware.com> nice to see koala up on a real web site :)
@1084978290 <andreas!andreas@cantor.gwydiondylan.org> Yes, indeed!
@1084978366 <andreas!andreas@cantor.gwydiondylan.org> It's pretty stable now. I've seen a very rare crash, but it seems to be an unhandled condition, not a runtime fault.
@1084978399 <andreas!andreas@cantor.gwydiondylan.org> "very rare" means running for an hour, serving 50 requests per second.
@1084978437 <andreas!andreas@cantor.gwydiondylan.org> There's one other problem: if there are too many concurrent threads (more than 10 or so), the garbage collector becomes *really* slow.
@1084979229 <cgay!~cgay@ita4fw1.itasoftware.com> awesome.  i'm going to rebuild with your network changes
@1084979301 <cgay!~cgay@ita4fw1.itasoftware.com> nice body painting, by the way :)
@1084979493 <andreas!andreas@cantor.gwydiondylan.org> Heh, thanks!
@1084979522 <andreas!andreas@cantor.gwydiondylan.org> I think you will need some more changes, especially regarding the MPS integration.
@1084979594 [hannes!~hannes@port-195-158-171-122.dynamic.qsc.de] ACTION still has only a minimal-console-compiler with mps which allocates memory and does nothing :(
@1084979726 <cgay!~cgay@ita4fw1.itasoftware.com> andreas, could you update the README.txt with the latest instructions?
@1084979828 <cgay!~cgay@ita4fw1.itasoftware.com> (when you get a chance)
@1084980747 <andreas!andreas@cantor.gwydiondylan.org> I will. But I also need to unroll about half a dozen workarounds, write bug reports, commit changes, etc.
@1084980962 <cgay!~cgay@ita4fw1.itasoftware.com> ok
@1084983260 -dircproxy- You disconnected
@1084985483 -dircproxy- You connected
@1084985622 <housel!housel@cantor.gwydiondylan.org> for some of the changes, we need to figure out "why" before committing to them
@1084985752 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> anybody seen this: " -                       Linking shared-object dfmc-modeling.so\n Killed" during phase 1 bootstrap (trying to get mps working with fundev, mps tests run ok).
@1084985806 <housel!housel@cantor.gwydiondylan.org> which platform?
@1084985819 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> FreeBSD in linuxulator
@1084985948 <housel!housel@cantor.gwydiondylan.org> do you have a core dump?
@1084986044 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> no, can't find a core file
@1084986113 <housel!housel@cantor.gwydiondylan.org> hmmm
@1084986216 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> it tries to link minimal-console-compiler.
@1084986383 <housel!housel@cantor.gwydiondylan.org> Phase 1 doesn't involve MPS, everything uses the 2.1alpha3 fdcompile
@1084986452 <housel!housel@cantor.gwydiondylan.org> I haven't done a Phase 1 lately, let me try it
@1084986511 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1084987495 <housel!housel@cantor.gwydiondylan.org> I'm getting weird responses from the CVS server lately
@1084987507 <housel!housel@cantor.gwydiondylan.org> "nothing known about Samples"
@1084987518 <housel!housel@cantor.gwydiondylan.org> "nothing known about admin"
@1084988877 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hmm, running fdcompile with strace prints the following before dying: fork()                                  = 7803
@1084988877 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> sendmsg?(673047680, {msg_name(-1010172835)={sa_family=0x75 /* AF_??? */, sa_data="\213E\10\203\370\1\213U\f\213u\20\17\204"}, msg_iov(1347675529)=No memory, msg_controllen=3280034560, msg_control=0xe8, , msg_flags=MSG_PEEK|MSG_CTRUNC|MSG_DONTWAIT|0x7c400}, MSG_OOB|MSG_PEEK|MSG_CTRUNC|MSG_EOR|MSG_WAITALL|MSG_TRUNC|0x1e00 <unfinished ...>
@1084990032 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1084990034 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> re
@1084990499 <cpage!~Chris@67.118.119.204> cgay: That's very cool, seeing Koala running.
@1084990861 <cgay!~cgay@ita4fw1.itasoftware.com> that's andreas' work, but yes!
@1084990886 -zelazny.freenode.net- hannes (~hannes@pD9E79E7A.dip.t-dialin.net) joined the channel
@1084990889 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> re
@1084990921 <cgay!~cgay@ita4fw1.itasoftware.com> hi hannes
@1084990924 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> hi carl
@1084991926 <housel!housel@cantor.gwydiondylan.org> does bsd strace work for linuxulator binaries?
@1084992020 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> not sure about that
@1084992208 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> btw, anyone updated cvs on cantor? there is a remote heap buffer overflow in pserver
@1084993429 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1084993802 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> hmm, stage1 works on a linux box.
@1084993850 <housel!housel@cantor.gwydiondylan.org> the whole thing works fine for me... I'm in the middle of a Stage3 (final)
@1084993878 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> i'm trying with a fresh cvs repository on my laptop
@1084993937 [hannes!~hannes@pD9E79E7A.dip.t-dialin.net] ACTION has no idea why it is broken :(
@1084999567 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> hmm, with a fresh installed functional-developer rpm and a freshly checked out fundev tree i can't bootstrap on FreeBSD-current :(
@1084999601 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Should fundev bootstrap on BSD?
@1084999608 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Indeed should it boostrap at all?
@1084999629 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> robmyers: it did in linuxulator ~2 months ago ;)
@1084999645 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Ah.
@1084999654 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> robmyers: bootstrapping with functional-developer-linux-alpha-rpm i've installed
@1084999699 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Where does it fail? GC, build system?
@1084999728 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> it fails during linking and says only 'Killed' not by whom.
@1084999751 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> ..linking minimal-console-compiler
@1084999797 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Can you gdb that step of the build? (says the UNIX expert ;-) )
@1084999867 <housel!housel@cantor.gwydiondylan.org> it did in linuxlator ~2 hours ago
@1084999900 <housel!housel@cantor.gwydiondylan.org> "Killed" often means "out of swap space"
@1084999909 <housel!housel@cantor.gwydiondylan.org> anything in dmesg?
@1084999914 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> housel: i've 512m ram
@1084999946 <housel!housel@cantor.gwydiondylan.org> resource limits? ("unlimit")
@1084999950 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> housel: ah, yes, i removed my swap partition, so it needs more than 512m ram
@1085000001 <housel!housel@cantor.gwydiondylan.org> why would you want to remove your swap partition?
@1085000041 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> due security reasons: it can contain private information when anybody gets access to my laptop.
@1085000061 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> and i thought 512m ram would be enough for me ;)
@1085000082 <housel!housel@cantor.gwydiondylan.org> can't you swap on an encrypted swap partition using bde?
@1085000125 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> yes, i could. i already use bde for my crypto filesystems (my ~) ;)
@1085000252 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> and i'd have no chance to get kernel dumps on a raw device (you can't dump to an encrypted swap partition and restore it)
@1085000720 <housel!housel@cantor.gwydiondylan.org> that's ok, I've gotten maybe 2 useful kernel dumps in the 7 years I've been running FreeBSD
@1085000776 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> housel: yes, this functionality would be a really nice feature if it would work ;)
@1085001599 <housel!housel@cantor.gwydiondylan.org> so does it work when you enable swap space?
@1085001634 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> yes, just finished Bootstrap.1 :)
@1085001808 <cpage!~Chris@67.118.119.204> re
@1085002434 <housel!housel@cantor.gwydiondylan.org> hi chris
@1085002479 <housel!housel@cantor.gwydiondylan.org> andreas: I think koala needs to have "automatic-finalization-enabled?() := #t;" before it will do the finalization
@1085002610 <housel!housel@cantor.gwydiondylan.org> that creates an extra thread that just sits around and waits for finalization messages from MPS
@1085003690 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> hmm, should we mention that > 512MB ram is needed for fundev bootstrapping ;)
@1085003790 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> s/;)/?/
@1085003802 <housel!housel@cantor.gwydiondylan.org> dunno
@1085003812 <cgay!~cgay@ita4fw1.itasoftware.com> s/ram/VM   ?
@1085003864 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> cgay: yes.
@1085005720 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> ok, stage2 finished, but minimal-console-compiler from stage2 doesn't run
@1085005821 <lisppaste!~lisppaste@69.55.225.112> hannes pasted "gdb backtrace of running stage2 minimal-console-compiler" at http://www.common-lisp.net/paste/display/1000
@1085006184 <bhoult!~bhoult@morannon.eservglobal.co.nz> eek.  512 MB!
@1085006196 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION remembers building d2c on a 32 MB machine.
@1085006210 <bhoult!~bhoult@morannon.eservglobal.co.nz> OK, that was painful, but 128 MB is plenty :-)
@1085006235 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> bhoult: with 512MB it fails to compile, it needs more.
@1085006264 <bhoult!~bhoult@morannon.eservglobal.co.nz> more total address space, yes I see that now
@1085006290 <bhoult!~bhoult@morannon.eservglobal.co.nz> but does 512 physical + swap perform OK (for those with swap :-) ??
@1085006366 <hannes!~hannes@pD9E79E7A.dip.t-dialin.net> bhoult: yes. it needs 530 MB vm
@1085006699 -zelazny.freenode.net- hannes_ (~hannes@pD9E7B019.dip.t-dialin.net) joined the channel
@1085006701 <hannes_!~hannes@pD9E7B019.dip.t-dialin.net> re
@1085007007 [prom!~prom@217.13.206.33] ACTION once tried to mindy-boot on an 80mb hppa box. but it was just too slow. i gave up after a week or so into the run when i wanted to move the machine...
@1085007190 <housel!housel@cantor.gwydiondylan.org> hannes: could you tell me the instruction at eip in frame #10?
@1085007342 <housel!housel@cantor.gwydiondylan.org> (I suspect I know already)
@1085007547 <housel!housel@cantor.gwydiondylan.org> also the values of src and dest in frame #0
@1085007554 <hannes_!~hannes@pD9E7B019.dip.t-dialin.net> uhm, what was the gdb command to show the instruction?
@1085007943 <housel!housel@cantor.gwydiondylan.org> disassemble
@1085007990 <hannes_!~hannes@pD9E7B019.dip.t-dialin.net> yes, found it
@1085008157 <bhoult!~bhoult@morannon.eservglobal.co.nz> must have been an old hppa.  The ones here aren't fast, but a mindystrap is only overnight
@1085008174 <bhoult!~bhoult@morannon.eservglobal.co.nz> HP-UX rand B.11.00 A 9000/785
@1085008183 <housel!housel@cantor.gwydiondylan.org> bhoult: the only reason it needs 512MB is that Phase 1 builds the entire compiler in one swell foop
@1085008192 <housel!housel@cantor.gwydiondylan.org> later phases do it library-at-a-time
@1085008233 <bhoult!~bhoult@morannon.eservglobal.co.nz> my std CPU speed test takes 4.4 seconds on the 9000/785
@1085008240 <lisppaste!~lisppaste@69.55.225.112> hannes annotated #1000 with "more info for peter" at http://www.common-lisp.net/paste/display/1000#1
@1085008242 <bhoult!~bhoult@morannon.eservglobal.co.nz> that's: time perl -e 'for($i=0;$i<=1000000;++$i){$t+=$i}print"$t\n"'
@1085008287 <bhoult!~bhoult@morannon.eservglobal.co.nz> that's just *slightly* faster than my 266 MHz G3 laptop
@1085008956 <hannes_!~hannes@pD9E7B019.dip.t-dialin.net> another try gives me: #0  0x28175c0a in __mprotect () at __mprotect:-1
@1085009016 <housel!housel@cantor.gwydiondylan.org> hmmm
@1085010618 -zelazny.freenode.net- hannes_ (~hannes@pD9E7B019.dip.t-dialin.net) joined the channel
@1085011482 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085012563 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085012565 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> re
@1085013618 <prom!~prom@217.13.206.33> bhoult: yeah. it was an old one. just a gecko. with a slow scsi hard disk, swapping all the time...
@1085013635 <prom!~prom@217.13.206.33> but not on hpux. i had linux on it.
@1085014000 <prom!~prom@217.13.206.33> (a gecko 60 or 80. 712/60 or 712/80. very small, light and nice workstation. but pretty slow ;) )
@1085014053 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1085015254 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1085015350 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) left the channel
@1085018585 -dircproxy- You disconnected
@1085023058 -dircproxy- You connected
@1085031435 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085031435 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085032003 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085032003 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085033406 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085033406 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085037050 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085037050 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085045568 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> morning
@1085048946 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085048946 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085051221 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085051258 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085062274 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> ok, fundev works with mps on freebsd if i comment out dylan exception handlers for sigfpe and sigsegv (in x86-linux-exceptions.c)
@1085066042 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/cgi-bin/bugzilla/show_bug.cgi?id=6202 is nasty
@1085066786 <housel!housel@cantor.gwydiondylan.org> probably best to fix it looking for all classes with repeated slots, and adding Dylan-level checks, rather than fixing it in collector.c
@1085067916 -dircproxy- You disconnected
@1085071594 -dircproxy- You connected
@1085072957 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1085073322 -dircproxy- You disconnected
@1085073447 -dircproxy- You connected
@1085074088 <housel!housel@cantor.gwydiondylan.org> actually maybe the Dylan-level layer (found in boot.dylan) that calls the primitive allocators would be even better
@1085074404 -dircproxy- You disconnected
@1085075074 -zelazny.freenode.net- hannes (~hannes@pD9E7B019.dip.t-dialin.net) joined the channel
@1085075118 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> re
@1085076968 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> while runnning koala and serving a directory listing i get MPS ASSERTION FAILURE: base != NULL (trace.c:1303). anyone seen this?
@1085082020 -dircproxy- You connected
@1085082370 <housel!housel@cantor.gwydiondylan.org> hannes: no, I haven't seen that particular problem
@1085082437 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> housel: hmm, it is not really easy to trace, cause FreeBSD gdb doesn't know anything about glibc and linux gdb is not fully able to read FreeBSD .core files :(
@1085082511 <housel!housel@cantor.gwydiondylan.org> I imagine it could figure out glibc with a little manual coaxing
@1085082529 <housel!housel@cantor.gwydiondylan.org> so far I haven't needed to know exactly which libc function I was in
@1085083171 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> housel: hmm, the FreeBSD gdb is also confused because there are signals he doesn't know
@1085083614 <housel!housel@cantor.gwydiondylan.org> when running MPS programs under gdb you have to do "handle SIGSEGV noprint nostop", or are you referring to some other signal?
@1085083639 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> i refer to "Program received signal ?, Unknown signal."
@1085083669 <housel!housel@cantor.gwydiondylan.org> oh, I think that's related to multithreading
@1085083685 <housel!housel@cantor.gwydiondylan.org> yes, I imagine it does get confused
@1085083765 <hannes!~hannes@pD9E7B019.dip.t-dialin.net> "warning: Signal ? does not exist on this system." is also a nice warning ;)
@1085087594 <cpage!~Chris@67.118.119.204> re
@1085088931 <housel!housel@cantor.gwydiondylan.org> hi chris
@1085089461 <bhoult!~bhoult@morannon.eservglobal.co.nz> morning
@1085089491 <bhoult!~bhoult@morannon.eservglobal.co.nz> so: I have a new machine at home, which came with Mandrake 10 on.  But there's a problem...
@1085089536 <bhoult!~bhoult@morannon.eservglobal.co.nz> gcc -static doesn't work for building (static) executables.  It fails because it can't find -ldl
@1085089551 <bhoult!~bhoult@morannon.eservglobal.co.nz> (and others, if I remove the -ldl from the link line)
@1085089585 <bhoult!~bhoult@morannon.eservglobal.co.nz> I see that Andreas had put "gcc -static" on a number of platforms about 9 months ago, and then abcked out all but x86-linux-gcc, saying that it worked there
@1085089714 <housel!housel@cantor.gwydiondylan.org> didn't work on FreeBSD either
@1085089801 <bhoult!~bhoult@morannon.eservglobal.co.nz> it works on some x86 linuxes
@1085089917 <bhoult!~bhoult@morannon.eservglobal.co.nz> or, at least, I assume it does
@1085089954 <bhoult!~bhoult@morannon.eservglobal.co.nz> it was left on x86-linux-gcc and x86-linux-gcc33, and l come to think of i I don't have one of those (I have gcc32)
@1085090240 [bhoult!~bhoult@morannon.eservglobal.co.nz] ACTION thinks that "gcc-Xlinker -Bstatic" does work
@1085090254 <bhoult!~bhoult@morannon.eservglobal.co.nz> or, at least well enough for our ICFP entries in 2000 and 2001
@1085090305 <housel!housel@cantor.gwydiondylan.org> I think I had trouble with that too, at least I always removed it at the time
@1085090323 <housel!housel@cantor.gwydiondylan.org> or -Wl,-Bstatic maybe it was
@1085090421 <bhoult!~bhoult@morannon.eservglobal.co.nz> well, gnu ld accepts -Bstatic or -dn or -non_shared or -static as the same meaning
@1085090464 <bhoult!~bhoult@morannon.eservglobal.co.nz> the -Wl, and -Xlinker are just variants for gcc, yes?
@1085090594 [housel!housel@cantor.gwydiondylan.org] ACTION checks
@1085090682 <housel!housel@cantor.gwydiondylan.org> looks like they're similar, -Wl splits args at commas, -Xlinker can be repeated
@1085090928 <bhoult!~bhoult@morannon.eservglobal.co.nz> hmm.  why do we have "/usr/local/lib/libgc.so: and not "-lgc" ??
@1085090951 <bhoult!~bhoult@morannon.eservglobal.co.nz> both work :-)
@1085090976 <bhoult!~bhoult@morannon.eservglobal.co.nz> but only the 2nd lets you statically link the gc
@1085090994 <bhoult!~bhoult@morannon.eservglobal.co.nz> so now "nm -u hello" is down to:
@1085091003 <bhoult!~bhoult@morannon.eservglobal.co.nz> _dl_hwcap
@1085091003 <bhoult!~bhoult@morannon.eservglobal.co.nz> dl_iterate_phdr
@1085091004 <bhoult!~bhoult@morannon.eservglobal.co.nz> _dl_rtld_map
@1085091006 <bhoult!~bhoult@morannon.eservglobal.co.nz> _DYNAMIC
@1085091008 <bhoult!~bhoult@morannon.eservglobal.co.nz> __gmon_start__
@1085091010 <bhoult!~bhoult@morannon.eservglobal.co.nz> _Jv_RegisterClasses
@1085091012 <bhoult!~bhoult@morannon.eservglobal.co.nz> __pthread_rwlock_rdlock
@1085091014 <bhoult!~bhoult@morannon.eservglobal.co.nz> __pthread_rwlock_unlock
@1085091053 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. but that's enough for "file hello" to say it's statically linked
@1085091142 <bhoult!~bhoult@morannon.eservglobal.co.nz> guess I'll raise a bug and then fix it ... saying "-L/usr/local/lib -lgc" should be more than satisfactory
@1085091162 <bhoult!~bhoult@morannon.eservglobal.co.nz> .. where the -L is the configured GC_PREFIX, of course
@1085092188 <bhoult!~bhoult@morannon.eservglobal.co.nz> Well, moving the -Xlinker -Bstatic to platforms.descr instead of the .lid file means that if something else works on another platform then that can be captured in the appropriate platforms.descr entry
@1085092217 <bhoult!~bhoult@morannon.eservglobal.co.nz> anyway I've raised bug 7021 for it
@1085093371 -zelazny.freenode.net- hannes_ (~hannes@pD9E7AA8E.dip.t-dialin.net) joined the channel
@1085097042 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085101731 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085105913 -zelazny.freenode.net- bhoult (~bhoult@morannon.eservglobal.co.nz) left the channel
@1085106219 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1085106299 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> hi
@1085106347 <housel!housel@cantor.gwydiondylan.org> hi carl
@1085106391 <housel!housel@cantor.gwydiondylan.org> I've been going through the Libraries/* bugzilla entries and figuring out which ones still apply
@1085106459 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> are a lot obsolete?
@1085106523 <housel!housel@cantor.gwydiondylan.org> not many... of 128 I closed maybe 8 or 9
@1085106560 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i've been wavering back and forth on whether to add file-system-case-sensitive? to the file-system module, or something like compare-filename to locators...
@1085106584 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i'm leaning toward the former because i can't think of a good name for the latter
@1085106595 <housel!housel@cantor.gwydiondylan.org> \=
@1085106669 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i'd have to do a ton of coersion to/from locators in koala to find out if index.html exists.  that's what i'd like to avoid
@1085106957 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i've never used OS X.  i just read something about it having an _optional_ case-sensitive file system.  do you know how that works?  i guess we would have to do something to support it in the system lib
@1085107113 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i would've thought it was all case-sensitive
@1085107159 <housel!housel@cantor.gwydiondylan.org> POSIX provides pathconf() which allows you to determine whether a particular path is case-sensitive or not
@1085107344 <housel!housel@cantor.gwydiondylan.org> no, I'm wrong, it does no such thing
@1085107408 <housel!housel@cantor.gwydiondylan.org> Anyway, the same thing can happen on Linux or BSD if a file is located on a VFAT filesystem
@1085107641 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> ok, i can just use \=.  i shouldn't be worrying about optimizing anything yet anyway
@1085108261 <housel!housel@cantor.gwydiondylan.org> speaking of optimizing, have you thought about the "pool of pre-initialized worker threads" idea?
@1085108976 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> not really.  i have a simple resource pool protocol that could probably be used.
@1085109136 [carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com] ACTION checks the logs...
@1085109318 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i can't tell from the logs if you guys thought it was required for the creator to call join()
@1085109392 <housel!housel@cantor.gwydiondylan.org> andreas figured out that it wasn't necessary, but he had to call release or something like that if join was not to be called
@1085109412 <housel!housel@cantor.gwydiondylan.org> so he called release in the thread finalizer primitive
@1085109505 <housel!housel@cantor.gwydiondylan.org> but then the finalizer wasn't being called
@1085109510 <housel!housel@cantor.gwydiondylan.org> so he called the finalizer manually
@1085109524 <housel!housel@cantor.gwydiondylan.org> then I figured out that he needed to start the automatic finalization thread
@1085109534 <housel!housel@cantor.gwydiondylan.org> and that's where it stands right now
@1085109665 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> what's the auto finalization thread? 
@1085109715 <housel!housel@cantor.gwydiondylan.org> that's what gets started when you do "automatic-finalization-enabled?() := #t;"
@1085109875 <housel!housel@cantor.gwydiondylan.org> it waits around for finalization events from MPS and calls finalize() when it receives them
@1085109966 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> and it's even documented :)
@1085109979 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i was gonna say...will it work under boehm?
@1085110023 <housel!housel@cantor.gwydiondylan.org> I don't think the current FunDev-Boehm integration supports finalization
@1085110053 <housel!housel@cantor.gwydiondylan.org> it has its own way of doing finalizers, but I don't think that way is being used right now
@1085110150 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i c.  at any rate, koala should use a thread pool.  and if not, it should at least call join
@1085114686 <cpage!~Chris@67.118.119.204> The HFS+ file system, which is the default on Mac OS X, is case preserving, but insensitive...just like Dylan :-)
@1085114716 <cpage!~Chris@67.118.119.204> You can mount case-sensitive volumes, of course.
@1085114728 <cpage!~Chris@67.118.119.204> You can optionally format the root volume using UFS, which is case-sensitive.
@1085114760 <cpage!~Chris@67.118.119.204> And more recently, Apple introduced an option to make an HFS+ volume that is case-sensitive, but it's for special cases.
@1085114819 <cpage!~Chris@67.118.119.204> Case-sensitive file systems still break some Mac applications, so Apple warns against using one for the root volume except for special cass.
@1085114824 <cpage!~Chris@67.118.119.204> s/cass/cases
@1085114851 <cpage!~Chris@67.118.119.204> And with that, I'm off to dinner.
@1085116589 -dircproxy- You disconnected
@1085116599 -dircproxy- You connected
@1085119807 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085131573 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1085131745 <andreas!andreas@cantor.gwydiondylan.org> Regarding the finalization: the finalization queue is manually darined on thread startup, so enabling automatic finalization shouldn't make muchof a difference.
@1085132538 <andreas!andreas@cantor.gwydiondylan.org> And yes, I've updated the CVS service.
@1085135886 <hannes!~hannes@port-195-158-171-122.dynamic.qsc.de> hi
@1085143225 <andreas!andreas@cantor.gwydiondylan.org> ~.
@1085144194 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085144215 <pritlove!~pritlove@port-212-202-175-14.dynamic.qsc.de> http://www.oreilly.com/news/graphics/prog_lang_poster.pdf
@1085145226 -zelazny.freenode.net- hannes (~hannes@pD9E7AA8E.dip.t-dialin.net) joined the channel
@1085145229 <hannes!~hannes@pD9E7AA8E.dip.t-dialin.net> re
@1085146054 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1085146529 -zelazny.freenode.net- pritlove_ (~pritlove@pD9E7AA8E.dip.t-dialin.net) joined the channel
@1085150244 <housel!housel@cantor.gwydiondylan.org> its lisp history is pretty lame
@1085150515 <andreas!andreas@cantor.gwydiondylan.org> Hi Peter!
@1085150790 <housel!housel@cantor.gwydiondylan.org> morning
@1085151605 -zelazny.freenode.net- pritlove__ (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085151668 <housel!housel@cantor.gwydiondylan.org> I still think it would make sense to add a joinable?: init-keyword to <thread>
@1085152231 -zelazny.freenode.net- pritlove (~pritlove@pD9E7AA8E.dip.t-dialin.net) joined the channel
@1085152806 <andreas!andreas@cantor.gwydiondylan.org> Probably.
@1085154311 -zelazny.freenode.net- pritlove (~pritlove@pD9E7AA8E.dip.t-dialin.net) joined the channel
@1085155332 -dircproxy- You disconnected
@1085155380 -dircproxy- You connected
@1085159621 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1085165103 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1085165106 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> heya
@1085165231 <housel!housel@cantor.gwydiondylan.org> I'm wondering how to solve the "multiple args to join" problem.  I'm thinking maybe we could give up on calling pthread_join entirely, and have a condition variable for thread termination that would replace it
@1085165296 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Can't you just pass a big fat object in through the pointer?
@1085165392 <housel!housel@cantor.gwydiondylan.org> the problem is that pthread_join only lets you wait for one particular thread to terminate, and Dylan join wants to be able to wait for one-of-n particular threads
@1085165489 <housel!housel@cantor.gwydiondylan.org> http://www.functionalobjects.com/products/doc/core/core_133.htm#HEADING133-0
@1085166007 <housel!housel@cantor.gwydiondylan.org> fixing this particular limitation is probably not a high priority, though
@1085166502 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Can you terminate them asynchronously and have them notify something as their final act?
@1085166512 <housel!housel@cantor.gwydiondylan.org> yes, that's the idea
@1085166543 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Been doing threads at work this week. Oh the joys of mapping from Win32 to Posix. :-)
@1085166633 <housel!housel@cantor.gwydiondylan.org> indeed
@1085167541 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> You could always use a thread variable. :-)
@1085167695 <cpage!~Chris@67.118.119.204> re
@1085167869 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Heya chris
@1085167897 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Just looking at pictures of Lisp Machines hosted in Macs (MacII-era). Scary stuff.
@1085171084 <cpage!~Chris@67.118.119.204> Which pictures?
@1085173914 <prom!~prom@217.13.206.33> rob: andreas has one of these. a symbolics ivory nubus card.
@1085173956 [Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net] ACTION wants to get one of those.
@1085173969 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I think I'll look for one in the June flea at MIT.
@1085173990 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Of course, people probably wouldn't want to give them up...
@1085174023 <prom!~prom@217.13.206.33> riastradh: iirc you can still buy them for a painfull price from symbolics. but andreas seems to have gotten the last symbolics mouse-and-kbd-to-adb converter they had for sale.
@1085174023 <^self!~fn@66.209.66.105> indeed.
@1085174060 <prom!~prom@217.13.206.33> riastradh: without this you cant use the great, strange symbolics kbd. it doesnt even have cursor keys ;)
@1085174160 <prom!~prom@217.13.206.33> but it has circle, triangle and rectanglge keys. and, of course, the hyper super meta modifiers.
@1085174185 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> If someone has a NuBus Lisp machine card, there will probably also be a Symbolics keyboard.
@1085174217 <prom!~prom@217.13.206.33> riastradh: yes. but you cant buy the adapter from symbolics... so youll have to find one in the wild.
@1085174256 [prom!~prom@217.13.206.33] ACTION tried to use genera with xforwarding, but the pc keyboard sucked for it.
@1085174257 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ...on the slim chances that I find the card & the keyboard, I suspect that whatever other equipment is necessary will be findable.
@1085174309 <prom!~prom@217.13.206.33> riastradh: yeah. just an old mac. and memory cards for the ivory. but that should come with the cpu card.
@1085174388 <prom!~prom@217.13.206.33> there is, of course, opengenera for the alpha.
@1085174416 <prom!~prom@217.13.206.33> it might be easier to get an alpha with a lot of memory and disk than an ivory card.
@1085174444 <prom!~prom@217.13.206.33> it certainly is easier...
@1085174503 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Eh.  That wouldn't be nearly as cool.
@1085174544 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> And I already have plenty of old Macs; I have no Alpha already.
@1085174618 [prom!~prom@217.13.206.33] ACTION only has a multia. but there is an alpha standing behind the couch here that has opengenera installed. 
@1085174667 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> And I recently bought a computer for running an old Lisp, so I'll wait a bit more before buying yet another computer for an old Lisp.
@1085174761 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (where 'recently' = 'one month ago')
@1085174934 <prom!~prom@217.13.206.33> riastradh: what kind of machine was thatß
@1085174937 <prom!~prom@217.13.206.33> ? even
@1085174955 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> A SPARC.
@1085175174 <prom!~prom@217.13.206.33> this reminds me of the situation on cantor and the fact that it wont be around for long.
@1085175213 <prom!~prom@217.13.206.33> the only alternative i can provide is a sun enterprise 450 with 2 CPUS @ 300 MHz and 1.7G ram.
@1085175248 <prom!~prom@217.13.206.33> of course, fd wouldnt run on that without the c backend running...
@1085175267 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (The old Lisp was T, whose linker I'm hacking now.)
@1085176213 <housel!housel@cantor.gwydiondylan.org> that would be a good incentive to get the C backend running again
@1085176228 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Just make a SPARC port!
@1085176257 <housel!housel@cantor.gwydiondylan.org> of course, but that can come later
@1085176280 <housel!housel@cantor.gwydiondylan.org> once we're sure Sun will survive :-)
@1085176281 <^self!~fn@66.209.66.105> you don't say.
@1085176284 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> How soon do you suppose a working PPC port will be ready for serious use?
@1085176330 <housel!housel@cantor.gwydiondylan.org> it wouldn't take very long if someone were to start seriously hacking on it
@1085176353 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> What would it take to convince one of you to do so?
@1085176371 <housel!housel@cantor.gwydiondylan.org> buy me a 12" iMac :-)
@1085176382 <cpage!~Chris@67.118.119.204> I'm up for it, but all I currently have are PPC machines, so I either need to get an x86 or VirtualPC.
@1085176400 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I find it hard to believe you can't get a shell on an x86 box anywhere.
@1085176455 <cpage!~Chris@67.118.119.204> Me? I suppose, but I have a hard time imagining that being a pleasant development environment.
@1085176499 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Emacs works fine remotely for me on my SPARC.  (I got a broken keyboard and haven't gotten around to getting a new one and setting up the windowing environment.)
@1085176512 <cpage!~Chris@67.118.119.204> In fact, offhand I don't know of one. Who wants to volunteer me one? :-)
@1085176565 [cpage!~Chris@67.118.119.204] ACTION does not consider emacs via ssh "pleasant," despite using emacs quite a lot recently and contributing to dylan-mode :-)
@1085176608 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Well, I guess it's more pleasant with T, since it's pretty much the same environment the original developers used, whereas with FunDev, there's a whole GUI development environment geared towards Windows...
@1085176626 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Or rather, not 'more pleasant,' but 'less comparatively unpleasant.'
@1085176929 <cpage!~Chris@67.118.119.204> Double-plus unpleasant?
@1085176961 <cpage!~Chris@67.118.119.204> There's the added pain of not having use of a meta key when going over ssh. Is there a way to work around having to type Esc separately?
@1085177058 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) joined the channel
@1085177137 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> cpage: emacs vi ssh doesn't float your boat? Right now, I'm dealing with an environment (complex and fragile CORBA-based system) that doesn't allow gdb debugging
@1085177146 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> its so barbarian
@1085177322 <cpage!~Chris@67.118.119.204> Ouch.
@1085177524 <rhashem!~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net> speaking of debugging --- does gdb work with fundev under Linux yet?
@1085177687 -zelazny.freenode.net- rhashem (~rhashem@dsl017-066-091.wdc1.dsl.speakeasy.net) left the channel
@1085178261 <housel!housel@cantor.gwydiondylan.org> cpage: you have a cantor account, right?
@1085178442 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ,exit
@1085178446 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Er, wrong window.
@1085178472 <housel!housel@cantor.gwydiondylan.org> What was that window? Scheme48?
@1085178478 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Yup.
@1085178623 <cpage!~Chris@67.118.119.204> housel: Yes.
@1085178880 <housel!housel@cantor.gwydiondylan.org> that should be sufficient I think
@1085179713 <cpage!~Chris@67.118.119.204> There are sufficient CPU and storage resources that builds wouldn't impact the web site or CVS usage?
@1085179999 <cpage!~Chris@67.118.119.204> Just had a thought: Does [] get turned into element() before or after the compiler looks at the arguments? The reason I ask is I'm wondering whether you could write a function macro named element() to mimic Perl's "list comprehension" syntax.
@1085180022 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Perl has a list comprehension syntax?
@1085180032 <cpage!~Chris@67.118.119.204> ...where you can write "[x**2 for x in rang(10)]" or "[x for x in S if x % 2 == 0]"
@1085180038 <cpage!~Chris@67.118.119.204> Sorry, I meant Python.
@1085180078 <housel!housel@cantor.gwydiondylan.org> you could write such a macro I imagine
@1085180081 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I think you should do something more like SRFI 42.
@1085180136 <cpage!~Chris@67.118.119.204> When I see this kind of thing in other languages, it makes me wish Dylan macros could define more fine-grained syntax, like operators.
@1085180150 <cpage!~Chris@67.118.119.204> ...or "expressions".
@1085180229 <cpage!~Chris@67.118.119.204> ie., I'm sure I could write a function macro, but I think the square bracket syntax is lighter-weight and stands out more than "foo(...)".
@1085180301 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> So don't write a function macro.
@1085180315 <cpage!~Chris@67.118.119.204> What's the alternative?
@1085180324 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Uh, the other kinds of macros.
@1085180334 <cpage!~Chris@67.118.119.204> ...which are heavier-weight than even function macros.
@1085180341 <cpage!~Chris@67.118.119.204> foo ... end;
@1085181283 [cpage!~Chris@67.118.119.204] ACTION takes a look at SRFI 42
@1085181352 <cpage!~Chris@67.118.119.204> BTW, why are for's "while:" and "until:" symbols, while "in", "then", etc. aren't?
@1085181387 <cpage!~Chris@67.118.119.204> Does that work around potential conflicts?
@1085181419 <housel!housel@cantor.gwydiondylan.org> yeah, probably
@1085181455 <cpage!~Chris@67.118.119.204> Any ideas about the kind of conflicts that could occur?
@1085181505 <housel!housel@cantor.gwydiondylan.org> while, at least, is already a statement-macro-word
@1085181584 <cpage!~Chris@67.118.119.204> Hmm. So, would "while" get parsed as a statement macro begin word in the context of such an hypothetical "for" macro?
@1085181628 <cpage!~Chris@67.118.119.204> I guess you want the start of the iteration clauses to allow arbitrary expressions?
@1085181644 <housel!housel@cantor.gwydiondylan.org> I'm not sure exactly, I haven't thought about it much
@1085181666 <cpage!~Chris@67.118.119.204> I'm wondering because this would be useful to know for macro writers.
@1085181698 <housel!housel@cantor.gwydiondylan.org> Perhaps it would be best to experiment
@1085181701 <cpage!~Chris@67.118.119.204> We could really use a more comprehensive macro tutorial.
@1085182114 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085182557 <housel!housel@cantor.gwydiondylan.org> I just killed off a bunch of tinderbox-related processes, cantor's load is going down to something reasonable
@1085185206 -zelazny.freenode.net- hannes (~hannes@62.206.114.179) joined the channel
@1085185209 <hannes!~hannes@62.206.114.179> re
@1085185345 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1085185369 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> buenas tardes
@1085185485 [housel!housel@cantor.gwydiondylan.org] ACTION just noticed that Yahoo maps has WiFi info
@1085185775 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1085185783 <hannes!~hannes@62.206.114.179> hi carl
@1085185933 <housel!housel@cantor.gwydiondylan.org> rhashem: in answer to your question, gdb works if all you want is a backtrace, and can live with mangled names
@1085185953 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> ok, thanks
@1085185968 <housel!housel@cantor.gwydiondylan.org> there's not been any debugger nub work for 2-3 weeks
@1085185981 <rhashem!~rhashem@ip68-100-5-48.dc.dc.cox.net> I think that's pretty much what I saw when I was trying to debug win32-duim under wine
@1085186806 -zelazny.freenode.net- cpage (~Chris@charger.intuit.com) joined the channel
@1085188917 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085189561 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085189606 -dircproxy- You disconnected
@1085189813 -zelazny.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1085189835 -zelazny.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) left the channel
@1085189872 -zelazny.freenode.net- cpage_ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1085191249 -dircproxy- You connected
@1085195441 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> erm, is there any way to get gdb to show me some local variables?
@1085195456 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> i guess the info just isn't there
@1085196166 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085196818 <housel!housel@cantor.gwydiondylan.org> no, it isn't
@1085196954 <housel!housel@cantor.gwydiondylan.org> right now HARP can only generate CodeView debugging info
@1085196998 <housel!housel@cantor.gwydiondylan.org> for ELF output we need DWARF2 debugging info
@1085197251 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> there's so much i don't understand in those two sentences, i don't know where to start.  :)
@1085197290 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> where does fundev/windows stand now?
@1085198053 <housel!housel@cantor.gwydiondylan.org> very reliable
@1085198065 <housel!housel@cantor.gwydiondylan.org> I'll try to do a release tomorrow
@1085198092 -zelazny.freenode.net- brucehoult (~bruce@202.0.59.8) joined the channel
@1085198161 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> cool.  g'night
@1085198882 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) left the channel
@1085199000 -zelazny.freenode.net- enometh (~Dadaism@adsl-66-159-224-83.dslextreme.com) joined the channel
@1085199419 -dircproxy- You disconnected
@1085199549 -zelazny.freenode.net- enometh (~Dadaism@adsl-66-159-224-83.dslextreme.com) left the channel
@1085207695 -zelazny.freenode.net- brucehoult_ (~bruce@202-0-60-23.adsl.paradise.net.nz) joined the channel
@1085208012 -dircproxy- You connected
@1085225477 -zelazny.freenode.net- lisppaste (~lisppaste@69.55.225.112) joined the channel
@1085225477 -zelazny.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1085225477 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085225477 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085225477 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085225477 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1085225477 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085225499 -zelazny.freenode.net- prom (~prom@217.13.206.33) joined the channel
@1085227601 -zelazny.freenode.net- robmyers (~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com) joined the channel
@1085232551 -zelazny.freenode.net- hannes (~hannes@62.206.114.179) joined the channel
@1085232554 <hannes!~hannes@62.206.114.179> re
@1085242169 -zelazny.freenode.net- pritlove (~pritlove@213.73.92.110) joined the channel
@1085242781 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> cpage: ping
@1085243922 <housel!housel@cantor.gwydiondylan.org> it's a little early for him perhaps
@1085244051 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Where is he?
@1085244261 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> I thought he was in the US?
@1085246810 <housel!housel@cantor.gwydiondylan.org> he's here on the west coast, but he seems to keep later hours than I do
@1085251464 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Yes, how I can chat to him sunday lunchtimes is beyond me. :-)
@1085253691 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> OK, he really ought to be up by now.
@1085253692 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> cpage: ping
@1085253762 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Maybe if you get him some coffee...
@1085253792 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Well, I can't DCC him any if he's not at his IRC client ready to receive it, so that plan won't work.
@1085257924 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Riastradh: pong
@1085258008 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Terrible latency there... :-)
@1085258092 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sorry, I don't always go first thing to the computer on Saturdays. ;-)
@1085260457 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085260629 -zelazny.freenode.net- brucehoult (~bruce@202-0-60-175.adsl.paradise.net.nz) joined the channel
@1085262620 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> cpage, did you look at SRFI 42 in much depth?
@1085263491 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Just a quick look. I've still got it open in my browser, though.
@1085263686 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> What do you think of it so far?
@1085263759 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Looks useful. Syntax isn't as nice as Python's.
@1085263797 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm thinking about implementing it for Dylan.
@1085263806 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I'm thinking about how to improve it.
@1085263835 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> And how to implement modularly it without syntax-rules -- its current implementation depends on CPS syntax-rules macros --.
@1085263884 <housel!housel@cantor.gwydiondylan.org> you'd rather use ?
@1085263910 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ?
@1085264144 <CIA-2!micah@hyperreal.info> commit by housel to fundev: Bug: 7023
@1085264144 <CIA-2!micah@hyperreal.info> Remove this binary; it can be built using open-source MPS.
@1085264274 <housel!housel@cantor.gwydiondylan.org> you want to use syntax-case, or standard syntax-rules, or what?
@1085264358 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Right now, with T's macro system, which is essentially defmacro.
@1085264405 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (I'm not exactly a fan of syntax-case, by the way, to put it, ah, lightly.)
@1085264777 <housel!housel@cantor.gwydiondylan.org> if you've read http://ftp.cs.indiana.edu/pub/scheme-repository/doc/misc/macros-*.txt I'm sure you're aware that I am a fan of syntax-case
@1085264819 <housel!housel@cantor.gwydiondylan.org> (11 years... where has the time gone?)
@1085264833 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Over there.
@1085264836 [Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net] ACTION gestures vaguely.
@1085264838 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> bad URL?
@1085264852 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> robmyers, no, you're just supposed to be a shell.
@1085264859 <housel!housel@cantor.gwydiondylan.org> strip off the filename, there's a glob
@1085264895 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Ahhhh. My globbing skills have gone to hell.
@1085264927 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> anyone know and love libtool?
@1085264942 <housel!housel@cantor.gwydiondylan.org> I know and tolerate libtool
@1085264945 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Mutually exclusive, surely. What's up?
@1085264964 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> I'm trying to figure out how to persuade d2c/libtool/gcc to make a statically linked executable on Mandrake 10
@1085265008 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> the stuff in platforms.descr that says "gcc -static" doesn't work, because it can't find -ldl -lpthreads -lm
@1085265040 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> they either don't *have* a .a lib version, or else it's named differently!
@1085265074 <housel!housel@cantor.gwydiondylan.org> I don't think libtool worries about "fully static"... it just makes sure what you compile with it is static, if you want statically linked libc etc you have to go outside it
@1085265080 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> saying "libtool --mode=link gcc -static" works, but still dynamically links against the above
@1085265098 <housel!housel@cantor.gwydiondylan.org> I suspect that libdl and libpthreads really don't have static versions
@1085265104 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Is the static lib a link to the dynamic (shudder)?
@1085265123 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> well, it also has a -all-static flag, which then expands to "gcc -static", which doesn't work
@1085265160 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> on Redhat8 it does work, sufficiently to persuade "file" that it is a statically linked program
@1085265200 <housel!housel@cantor.gwydiondylan.org> could those static libs be in some RPM you haven't installed yet?
@1085265220 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> possible, i guess.  I didn't do the OS install, the vendor did
@1085265261 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> it's pleasantly quick, btw :-)  6:40 for a d2c regular build
@1085265262 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Why doesn't gcc -static work when expanded but ok when in a libtool call? (I f I've read correctly).
@1085265310 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> when you do "libtool --mode=link gcc -static ..." libtool removes the -static and replaces various things e.g. -lgc with the full path to the .a file
@1085265383 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> there is e.g. a libpthread.so
@1085265391 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> and a libpthread_unshared.a
@1085265470 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> but it seems that libtool doesnt know to look for the _unshared name
@1085265557 <housel!housel@cantor.gwydiondylan.org> I wouldn't know to look for that either...
@1085265595 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> er, "nonshared"
@1085265599 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> sorry
@1085265605 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> but I wouldn't know either
@1085265607 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com>  make a link from libpthread.a to libpthread_unshared.a? (hack, hack)
@1085265624 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> there's a libc_nonshared.a as well
@1085265661 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> rob: yeah I know, that might well work, but I'd like to set up gwydion so it works out of the box on similar systems
@1085265672 [brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz] ACTION might just nuke it all and install debian...
@1085265689 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> housel, what don't you like about syntactic closures?
@1085265841 <housel!housel@cantor.gwydiondylan.org> I don't have anything against them
@1085265856 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Oh, OK.
@1085265906 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> My preference is either syntactic closures or explicit renaming, with destructuring syntax at the Scheme level.  (Of course, syntax-rules is fine too, for most macros.)
@1085266140 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> interesting
@1085266153 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> brucehoult: set up a *local* link for the lib as part of the config, or change the library name searched for on Mandrake?
@1085266164 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> the -ldl is neded for Boehm.  but when you statically link to gc then you don't need -ldl
@1085266166 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> so that's ok
@1085266197 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> .. annnnd, if you leave -lpthread off the command then libtool puts it back in!
@1085266231 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> ... annnnd, if you put -lpthread_nonshared instead then libtool puts -lpthread -lpthread_nonshared  :-(
@1085266232 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Is that the locally made libtool as in "making libtool"? Where's the config for that?
@1085266302 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> I'm trying both the libtool that came with the system (/usr/bin/libtool) and one I've built myself (/usr/local/bin/libtool).  no difference
@1085266323 <housel!housel@cantor.gwydiondylan.org> how about the one that was built when you ran configure?
@1085266337 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> d2c's one?
@1085266346 <housel!housel@cantor.gwydiondylan.org> ja
@1085266347 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> that's only used when you build d2c, right?
@1085266356 <housel!housel@cantor.gwydiondylan.org> ja
@1085266357 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> This is the one I was meaning. :-)
@1085266361 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> and it's not installed anywhere?
@1085266370 <housel!housel@cantor.gwydiondylan.org> richtig
@1085266373 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> I'm not trying to build, d2c, but to build hello-world
@1085266425 -zelazny.freenode.net- oudeis (~oudeis@bzq-218-95-152.red.bezeqint.net) joined the channel
@1085266430 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> in any case, it's the same results :-)
@1085266437 <housel!housel@cantor.gwydiondylan.org> hmmm
@1085266485 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hack the Makefile to get libtool to just run --version?
@1085266536 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> why?
@1085266546 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> I'm running libtool manually at the moment anyway
@1085266558 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz>  /usr/bin/libtool is 1.4.3
@1085266577 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> hello. anyone here with exper ience with fun-o functional developer?
@1085266579 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz>  /usr/local/bin/libtool and gwydion's one are both 1.5.6
@1085266618 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Oh, OK, just to make sure: used to screw up builds on MacOSX if libtool rather than glibtool got called, version was the only way to make sure. Thought version might illuminate here....
@1085266680 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> so cantor, as you migth expect, has libpthread.a and libpthread.so
@1085266687 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> .. and libc.a and libc.so
@1085266689 <housel!housel@cantor.gwydiondylan.org> oudeis: yes, I'm building a Functional Developer release right now
@1085266697 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> none of this _nonshared crap
@1085266744 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> ROFL; someone's self-description on Amazon.com: (somebody ((who) likes) LISP)
@1085266760 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> that's not valid CL
@1085266764 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> needs a funcall
@1085266802 <robmyers!~robmyers@spc1-pete2-5-0-cust77.lond.broadband.ntl.com> Hey, they don't say it's CL. :-)
@1085266809 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> well, unless "somebody" is a macro, I guess
@1085266828 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: i've just installed the trial version and am having problems with building standalone-deuce example. it doesn't find rc.exe to compile the resource files. i've copied rc.exe from my visual studio installation on another machine to dylan's bin directory, but it din't help. do you have an idea what might be the problem?
@1085266897 <housel!housel@cantor.gwydiondylan.org> hmmm... that's always worked for me, provided I had rc.exe (and the rest of Visual Studio) in my path
@1085266948 <housel!housel@cantor.gwydiondylan.org> let me try it on the release I built yesterday
@1085266975 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i don't have visual studio installed on this machine. but i've put fun-o bin on the path
@1085267021 <housel!housel@cantor.gwydiondylan.org> You also need link.exe from Visual Studio, and to choose "Microsoft Linker" in the Environment Options
@1085267066 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ah, yes, it says in the build log it's using gnu linker
@1085267066 <housel!housel@cantor.gwydiondylan.org> and to have the MSLinker library pack installed
@1085267068 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> housel: I think perhaps oudeis isn't aware of some significant news
@1085267117 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i'm not aware of any news: it's my day 0 with dylan :)
@1085267129 <housel!housel@cantor.gwydiondylan.org> yes: be sure to check http://www.gwydiondylan.org/~housel/fundev-snapshot.html later today, or tomorrow
@1085267191 <housel!housel@cantor.gwydiondylan.org> (don't bother downloading the snapshot that's there right now, it is known to crash occasionally)
@1085267402 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> wow
@1085267431 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> does this mean gwydion group is in charge of it now and it is actively developed again?
@1085267462 <housel!housel@cantor.gwydiondylan.org> yes
@1085267483 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> and it is going completely open-source?
@1085267509 <housel!housel@cantor.gwydiondylan.org> exactly
@1085267574 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> somebody should put a warning sign on the functional objects site - someone might shell USD400 for it in the meantime
@1085267608 <housel!housel@cantor.gwydiondylan.org> there's one there already... if you try to buy something it says the store is closed or something like that
@1085267636 <housel!housel@cantor.gwydiondylan.org> Go to Help... About and click the Purchase License button
@1085267647 <housel!housel@cantor.gwydiondylan.org> Register button, sorry
@1085267670 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> it is so good this excellent thing don't just die like many others...
@1085267690 <housel!housel@cantor.gwydiondylan.org> yes, that's how we feel
@1085267762 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> btw i'm a fairly proficient c++ programmer. if you need any (for now light) help, i am willing to participate
@1085267783 <housel!housel@cantor.gwydiondylan.org> great!
@1085267847 -zelazny.freenode.net- brucehoult (~bruce@202-0-60-175.adsl.paradise.net.nz) joined the channel
@1085267851 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i also have some background in cl, if that matters
@1085267892 <housel!housel@cantor.gwydiondylan.org> that's a big plus
@1085268075 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> gwydion isn't exactly "in charge" of Fun-Dev now
@1085268084 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> we have it in our cvs, and can do anything we like with it, true
@1085268119 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> .. but Fun-O are still planning to sell it if you want an actual CD etc
@1085268159 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> I don't know what would happen if we made changes to it that the Fun-O guys didn't like :-)
@1085268170 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> but they don't do any development?
@1085268197 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> well, they *can*, of course, but ... how much have they done the last few years? :-(
@1085268215 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> your relationship with them is governed by some kind of contract?
@1085268230 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> only the source license
@1085268238 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> .. and goodwill
@1085268324 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well people, if you think i might do something useful for starters in the next 2 hours, tell me
@1085268355 <housel!housel@cantor.gwydiondylan.org> I submitted bug#6986 for their approval and comment but got no response
@1085268376 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> where's the bug database?
@1085268406 <housel!housel@cantor.gwydiondylan.org> http://www.gwydiondylan.org/cgi-bin/bugzilla/index.cgi
@1085268412 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> recent comments from a Fun-O guy:
@1085268424 <brucehoult!~bruce@202-0-60-175.adsl.paradise.net.nz> [cgay] i'm just glad someone is working on it finally.  i wish i had gotten started earlier on open sourcing
@1085268458 <housel!housel@cantor.gwydiondylan.org> Search for Product="FunDev"
@1085268484 <housel!housel@cantor.gwydiondylan.org> we've imported their old interanal bug database
@1085268553 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> this search string doesn't bring anything up
@1085268617 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok got it hte other route
@1085268673 <housel!housel@cantor.gwydiondylan.org> currently 1484 bugs
@1085268779 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> who are fun-o now? is scott mckay still there?
@1085268780 -zelazny.freenode.net- brucehoult_ (~bruce@202-0-60-85.adsl.paradise.net.nz) joined the channel
@1085268818 <housel!housel@cantor.gwydiondylan.org> Scott McKay is at ITA software, the people who wrote the Orbitz travel search system
@1085268818 <^self!~fn@66.209.66.105> word.
@1085268869 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> run away, little bot
@1085268894 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i mean are they just 2-3 management guys now, holding the rights to ip?
@1085268926 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> they are (some of) the people who wrote it, at Harlequin
@1085268931 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> but they all have day jobs
@1085269000 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> and what is your development model? is it structured in any way, or you prefer not to think of these matters now?
@1085269047 <housel!housel@cantor.gwydiondylan.org> we each work on whatever interests us or seems urgent
@1085269058 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> like all open-source projects, we all work on what interests us most
@1085269065 <housel!housel@cantor.gwydiondylan.org> we prefer to center work around Bugzilla entries
@1085269091 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> are you the antipodes who took 2nd place on last year icfp contest?
@1085269099 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> that doesn't necessarily get the right stuff done, but there's no stick to make people work on stuff they don't want to do
@1085269099 <housel!housel@cantor.gwydiondylan.org> We're trying to institute more review, where the developer posts patches to the bugzilla entry before committing them
@1085269109 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> I am
@1085269118 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> and it's was Judges' Prize
@1085269143 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> myself, working with two friends here in Wellington, NZ
@1085269176 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> some other of the Gwydion people (Andreas and I put in the most time) got 2nd place in ICFP in 2001
@1085269216 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i think i mean that one, with race tracks
@1085269232 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> race tracks was last year.  And judges' Prize.
@1085269242 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> kinda a 3rd place, except we were actually 5th :-)
@1085269254 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> we beat the winner on 4 of the 9 tracks :-)
@1085269267 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> ... but really really sucked on one of them
@1085269279 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok. got it sorted out. i've read your account of the development
@1085269307 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> next contest is in 2 weeks...
@1085269344 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> if i enter, it will be in cl ;)
@1085269345 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 2001 was the lame markup one, wasn't it?
@1085269351 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> right
@1085269364 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> 2002 was the robot game.  I rather liked that one; I wish I had participated.
@1085269393 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> we sucked in 2002.  there were too many cooks
@1085269401 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Cooks?
@1085269437 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> after our success in 2001, everyone and his dog wanted to be in the team in 2002
@1085269443 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> it was a disaster
@1085269450 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Ah.
@1085269478 -zelazny.freenode.net- hannes (~hannes@62.206.114.179) joined the channel
@1085269483 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> I got fed up after 24 hours and resigned from the team and started a new team with friends working in my house
@1085269500 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> not enough time, though, and we got only 35th
@1085269504 <hannes!~hannes@62.206.114.179> hi
@1085269513 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> the original team got 99th or 100th
@1085269533 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Do you know what rank Studio Ghibli (or however it's spelled) got?
@1085269542 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> no
@1085269576 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> who was that?
@1085269579 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> and what were they using?
@1085269593 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> It would be nice to see a strong Scheme and/or CL team
@1085269603 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> there were several of each last year, but none did very well
@1085269611 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> My brother, a friend of his (he showed up in this channel once or twice under the nick pknodle (and he still has my Dylan books, argh!)), and some guy from BU; they used Scheme.
@1085269624 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: is there a way to turn on color-coding in deuce?
@1085269680 <housel!housel@cantor.gwydiondylan.org> What kind? there are color-coded dispatch markers, but no color-coded keywords
@1085269694 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> color-coded keywords
@1085269713 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Last year the #scheme team didn't do very well (although I hear we got first place on one of the tracks), but instead of writing a path finder, we wrote a path figurer given an approximate sequence of points and a GUI app for picking the points.
@1085269724 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> I'm a little unhappy that we used some C++ last year.  It was necessary only because our guy who knew OpenGL hadn't done any in Dylan himself, though we have OpenGL bindings for d2c
@1085269770 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: do you think it might be straigtforward to implement?
@1085269780 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> I wonder what this year's contest will be about.
@1085269786 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> +1
@1085269799 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ?
@1085269806 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> will know soon enough
@1085269814 <housel!housel@cantor.gwydiondylan.org> not sure... I haven't spent much time looking at Deuce sources
@1085269828 <brucehoult_!~bruce@202-0-60-85.adsl.paradise.net.nz> I *hope* those with access to supercomputer clusters won't have a big advantage
@1085269838 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Too soon.  I haven't had time to manage to cross-compile Stalin to the PPC, in case we need it...
@1085269872 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: do you think it is a desirable feature to anyone but me?
@1085269913 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> we have color-coding in our emacs dylan-mode
@1085269932 <housel!housel@cantor.gwydiondylan.org> it probably is a desirable feature.  (I usually use emacs, and that's probably one of the reasons)
@1085269941 [brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz] ACTION rants about editors that emulate emacs in the trivial aspects, but not the ones that make emacs emacs
@1085270067 [Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net] ACTION rants about elisp.
@1085270110 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> so how do i grab the sources from the cvs?
@1085270120 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> for fun-dev?
@1085270124 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> yes
@1085270132 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> if i can...
@1085270135 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> same instructions as d2c, but different module name
@1085270157 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> http://www.gwydiondylan.org/cgi-bin/viewcvs.cgi
@1085270163 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> module is called "fundev"
@1085270187 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> http://www.gwydiondylan.org/downloading.phtml
@1085270215 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> so something like
@1085270226 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> cvs -d :pserver:anoncvs@cvs.gwydiondylan.org:/var/lib/cvs co fundev
@1085270259 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> prolly want a -z3 or soemthing in there too
@1085270279 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> .. and need to cvs login first, of course
@1085270291 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> cvs -d :pserver:anoncvs@cvs.gwydiondylan.org:/var/lib/cvs login
@1085270293 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> anoncvs
@1085270322 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> what cvs client do you use on windows?
@1085270344 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> well, I don't, but I expect I'd use cygwin's one
@1085270485 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> pardon my ignorance, what does -z3 do?
@1085270499 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> compression
@1085270519 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085270520 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 3 is an arbitrary level from 1 - 9
@1085270528 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> source code compresses well :-)
@1085270550 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> If you choose something too low, it's slow to download; if you choose something too high, you annoy nearly everyone here who is using the server at the time.
@1085270550 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well i'm on a fairly fast link anyway
@1085270569 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> so i won't annoy you :)
@1085270572 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> our server is a pretty fast machine, so high compression is ok
@1085270582 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> z5 is usually the highest I ever use.
@1085270586 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> (celery 1.2, which is better than most web servers)
@1085270698 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Hey, one of the ICFP 2004 contest organizers bought my t-shirt!
@1085270713 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ?
@1085270728 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> http://www.cafeshops.com/skicalc
@1085270736 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> and, why do you need Stalin?  Won't Chicken do?
@1085270737 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (the front of the shirt is uninteresting; see the back)
@1085270761 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> If I really need speed, I'll use Stalin; if not, we'll use Scheme48.  (We'll be using Scheme48 anyways.)
@1085270798 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> yeah but Stalin will take all 72 hours to compile the program!
@1085270814 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> your cvs server does not support module lists :(
@1085270828 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> Yes...that's why we'll use Scheme48 anyways, except at the very end.
@1085270845 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (I might also try it in parallel with T.)
@1085270926 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> huh?
@1085270932 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> ?
@1085270954 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> like "cvs co src fundev www", you mean?
@1085270977 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well i have no idea, but i use  a nice gui client
@1085270993 <Riastradh!~riastradh@pool-141-154-214-206.bos.east.verizon.net> (A couple quick benchmarks (admittedly not particularly accurate) demonstrated that Orbit is considerably faster, in both compilation speed and generated code, than Chicken.)
@1085271005 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> which has an option to fetch available modules from cvs server, if server provides the list in some file
@1085271022 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ah
@1085271029 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> no, we don't have that, i think
@1085271056 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> viewcvs/cvsweb can get the module list, but only because it has direct filesystem access to the repository
@1085271058 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> anyway, i just put "fundev" in the appropriate textbox and it is churning away right now
@1085271079 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> cvs is so simple I cna't see a reason to put a GUI on it
@1085271108 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ... and I'm a guy who jumped right onto the Mac way in 1984 when it came out
@1085271115 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well, when i was looking for a cvs client for windows, i could find only gui ones :)
@1085271148 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> this particular one is called tortoisecvs and it is integrated nicely with windows explorer
@1085271180 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> it even colors cvs-controlled folders in irritating green
@1085271213 <housel!housel@cantor.gwydiondylan.org> The DSP engineers at my work use WinCVS (and can't use it any other way)
@1085271229 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> the only cvs GUI I use is vc.el  :-)
@1085271352 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i think ishould have taken your advice about -z option. how big is it?
@1085271461 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> the fundev module repository is 294172 KB
@1085271477 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> a checkout should be smaller than that :-)
@1085271485 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> why?
@1085271493 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> becasue that has all versions
@1085271499 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ah
@1085271503 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> and some old files that you won't get
@1085271503 <housel!housel@cantor.gwydiondylan.org> Sources alone is 29M
@1085271577 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> you're using about 2% CPU.  feel free to try again with compression :-)
@1085271592 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> it should just carry on from where it got to
@1085271604 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> if it is 29M, it is just a few more minutes
@1085271674 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> but strangely, the checkout is  >50M already
@1085271692 <housel!housel@cantor.gwydiondylan.org> doc is another 26M
@1085271744 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ok, I'm just doing a local checkout of fundev
@1085271970 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> yep it's big :-)
@1085272021 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> bruce@cantor:~$ du -ks fundev/*
@1085272021 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 16      fundev/CVS
@1085272022 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 28      fundev/Functional Objects Library Public License.txt
@1085272022 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 28      fundev/GNU Lesser General Public License.txt
@1085272022 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 8       fundev/Makefile.in
@1085272022 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 4       fundev/README.txt
@1085272024 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 124     fundev/Samples
@1085272026 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 192     fundev/Setup.bmp
@1085272028 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 78128   fundev/Sources
@1085272030 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 1608    fundev/admin
@1085272032 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 4       fundev/autogen.sh
@1085272034 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 260     fundev/build
@1085272036 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 4       fundev/configure.ac
@1085272038 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 183708  fundev/doc
@1085272040 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 4492    fundev/experimental
@1085272042 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 2140    fundev/install
@1085272044 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 20      fundev/local
@1085272046 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 10320   fundev/old
@1085272048 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 132     fundev/project
@1085272050 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 2004    fundev/releases
@1085272052 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 2244    fundev/tools
@1085272054 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 3400    fundev/training
@1085272056 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz>  
@1085272075 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> where did you get those numbers from, Peter?
@1085272083 <housel!housel@cantor.gwydiondylan.org> dir/s on Win32
@1085272107 <housel!housel@cantor.gwydiondylan.org> but I misread the doc number
@1085272138 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> can you build from just fundev/Sources?
@1085272152 <housel!housel@cantor.gwydiondylan.org> no
@1085272165 <housel!housel@cantor.gwydiondylan.org> admin/ is needed too, and install on Win32
@1085272186 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> they're small
@1085272237 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> do you think we should rearrange our repository to put the gd stuff into a subdirectory?
@1085272259 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> .. like it used to be on berlin.ccc.de
@1085272266 <housel!housel@cantor.gwydiondylan.org> you mean other than "src"?
@1085272308 <housel!housel@cantor.gwydiondylan.org> I'm recommending that we switch to Subversion when 1.1 comes out, then we can rearrange things any way we want
@1085272352 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> hmm.  I guess most of it is actually compiler.neutral
@1085272371 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ref isn't though
@1085272411 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> are hd-sources and fd-sources still useful?
@1085272436 <housel!housel@cantor.gwydiondylan.org> no
@1085272461 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> they're duplicates of stuff in fundev?
@1085272465 <housel!housel@cantor.gwydiondylan.org> yes
@1085272628 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> we'll want to rationalize src/duim and fundev/Sources/duim at some point too, I guess
@1085272640 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> .. possibly into libraries
@1085272650 <housel!housel@cantor.gwydiondylan.org> possibly
@1085272676 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> and qa
@1085272783 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> re: putting proposed patches into bugs...
@1085272798 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> ... I'm doubtfull that's useful
@1085272816 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> unless we expect many of them to be rejected
@1085272825 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> sorry to bother you, i have just rerun cvs from the command line with -z9 option - how's server health? here on the client side it is suddenly very slow
@1085272829 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> I mean, OK, sure, for people without cvs write access
@1085272867 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz>  6274 pubcvs    17   0 52412  51M   864 R    32.3 20.4   1:01 cvs
@1085272867 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz>  6336 pubcvs     8   0 36588  35M  1024 S    16.0 14.3   0:47 cvs
@1085272880 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> I don't know why there are two of them
@1085272909 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> -z9 is probably a bit much
@1085272915 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> I use -z3 :-)
@1085272932 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> enough to get all the cheap compression (of which there is plenty)
@1085273054 <housel!housel@cantor.gwydiondylan.org> I've gotten strange errors when using -z9, and have lately throttled back down to -z5
@1085273459 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> cool: with a statically-linked (almost) d2c installed, I can build d2c in 6:12
@1085273515 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> that's about 40s difference between statically-linked and dynamic
@1085273570 [brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz] ACTION can't believe he put up with 30 min builds for so long...
@1085273718 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok i've got it - the final count is 242M. how do i build it? it is surely a trivial operation :)
@1085273747 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> well, as we got it from Fun-O, you basically couldn't build it
@1085273765 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> but housel has been working hard at it
@1085273787 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> Basically, I think the plan is a big public announcement when it builds easily
@1085273858 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> stat with fundev/README.txt :-)
@1085273868 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085274677 <lisppaste!~lisppaste@69.55.225.112> oudeis pasted "mps build fails" at http://www.common-lisp.net/paste/display/1038
@1085274708 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> any idea why?
@1085275688 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> anyone built it with VS7?
@1085276080 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well of course: static const char digit[16]="0123456789abcdef";
@1085276146 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> 17
@1085276166 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> got it already:)
@1085276227 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> strange vs6 compiler haven't got it
@1085276309 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> shall i submit bug report to ravenbrook?
@1085276485 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> also, to whoever wrote the readme:
@1085276596 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> there's a mistake in mps building directions for win32, in the copy incantation:
@1085276617 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> s/mmdw.a/mmdv.lib
@1085276636 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> s/mmdw.a/mmdw.lib
@1085276645 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> I'm sure they'd appreciate bug reports
@1085277814 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085278281 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085278286 <brucehoult!~bruce@202-0-60-85.adsl.paradise.net.nz> cpage: the meta key works just fine over ssh.  If it doesn't then it's a problem with your terminal, not ssh
@1085278498 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> pretty tough task to get to Nick Levine with the bug reports
@1085278533 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> his junk mail filter bounces *anything* you throw at it
@1085280783 -zelazny.freenode.net- brucehoult_ (~bruce@202-0-45-5.adsl.paradise.net.nz) joined the channel
@1085285891 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085286149 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085291157 -zelazny.freenode.net- cpage (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1085293021 <housel!housel@cantor.gwydiondylan.org> I'll finish my release tomorrow afternoon
@1085293031 <housel!housel@cantor.gwydiondylan.org> once I fix bug 7024
@1085293032 <brucehoult_!~bruce@202-0-45-5.adsl.paradise.net.nz> cool!
@1085297135 -zelazny.freenode.net- cgay (~cgay@ita4fw1.itasoftware.com) joined the channel
@1085297135 -zelazny.freenode.net- andreas (andreas@cantor.gwydiondylan.org) joined the channel
@1085301496 <CIA-2!micah@hyperreal.info> commit by bruce to src: Bug: 7021
@1085301496 <CIA-2!micah@hyperreal.info> Safer, but slightly less static, version for x86-gcc and x86-gcc33
@1085301782 <CIA-2!micah@hyperreal.info> commit by bruce to src: Bug: 7021
@1085301782 <CIA-2!micah@hyperreal.info> Merge Transaction 73 to gd25
@1085302243 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> How is the meta key typically transmitted over an ssh connection?
@1085302273 <brucehoult_!~bruce@202-0-45-5.adsl.paradise.net.nz> I don't know.  but it certainly works fine
@1085302305 <brucehoult!~bruce@202-0-45-5.adsl.paradise.net.nz> At work I ssh from linux to solaris and hpux and run emacs sessions up the gazoo
@1085302385 <brucehoult!~bruce@202-0-45-5.adsl.paradise.net.nz> are you using the OSX terminal program?
@1085302404 <brucehoult!~bruce@202-0-45-5.adsl.paradise.net.nz> I just tested it, sshing into both Linux and another OSX, and meta works just fine
@1085302424 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, it's working for me now. Other times I've tried emacs via ssh I've had to use an explicit Esc instead. Maybe Terminal is set up differently now.
@1085302433 <brucehoult!~bruce@202-0-45-5.adsl.paradise.net.nz> (it's option, rather than my prefered command, but you can't have everything)
@1085302512 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, that's a little annoying. The option key isn't as easy to use.
@1085312572 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085319841 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085320094 -zelazny.freenode.net- hannes (~hannes@62.206.114.179) joined the channel
@1085320934 -zelazny.freenode.net- hannes_ (~hannes@62.206.114.179) joined the channel
@1085325447 <CIA-2!micah@hyperreal.info> commit by cpage to src: Bug: 7025 7026
@1085325447 <CIA-2!micah@hyperreal.info> Added support for fontification of sealed domain definitions.
@1085325447 <CIA-2!micah@hyperreal.info> Added support for applying separate faces to the names of module
@1085325447 <CIA-2!micah@hyperreal.info> <...>
@1085335301 -zelazny.freenode.net- brucehoult (~bruce@202-0-45-5.adsl.paradise.net.nz) joined the channel
@1085335303 -zelazny.freenode.net- hannes__ (~hannes@pD9E7A3DD.dip.t-dialin.net) joined the channel
@1085335775 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085335775 -zelazny.freenode.net- lisppaste (~lisppaste@69.55.225.112) joined the channel
@1085336342 -zelazny.freenode.net- brucehoult (~bruce@202-0-45-5.adsl.paradise.net.nz) joined the channel
@1085336437 -zelazny.freenode.net- brucehoult (~bruce@202-0-45-5.adsl.paradise.net.nz) joined the channel
@1085336437 -zelazny.freenode.net- lisppaste (~lisppaste@69.55.225.112) joined the channel
@1085336437 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085336437 -zelazny.freenode.net- hannes__ (~hannes@pD9E7A3DD.dip.t-dialin.net) joined the channel
@1085336656 -zelazny.freenode.net- lisppaste (~lisppaste@69.55.225.112) joined the channel
@1085336656 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085336656 -zelazny.freenode.net- hannes__ (~hannes@pD9E7A3DD.dip.t-dialin.net) joined the channel
@1085337445 -zelazny.freenode.net- lisppaste (~lisppaste@69.55.225.112) joined the channel
@1085337445 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085337445 -zelazny.freenode.net- hannes (~hannes@pD9E7A3DD.dip.t-dialin.net) joined the channel
@1085346210 -zelazny.freenode.net- pritlove (~pritlove@pD9E7A3DD.dip.t-dialin.net) joined the channel
@1085348879 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> re
@1085349303 <housel!housel@cantor.gwydiondylan.org> brucehoult: the point of review isn't that the code might be rejected, it's to give other people opportunity to suggest improvements, or to point out style violations that can be fixed
@1085349387 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> housel: What's that in reply to?
@1085349445 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (I don't see anything in the log.)
@1085349791 <housel!housel@cantor.gwydiondylan.org> something he said yesterday
@1085349881 <housel!housel@cantor.gwydiondylan.org> <brucehoult> re: putting proposed patches into bugs...
@1085349881 <housel!housel@cantor.gwydiondylan.org> <brucehoult> ... I'm doubtfull that's useful
@1085349881 <housel!housel@cantor.gwydiondylan.org> <brucehoult> unless we expect many of them to be rejected
@1085349934 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'd only be concerned about the possibility of having a lot of large patches adding noise and storage requirements to the bug database. Would they be inline or attachments?
@1085350016 <housel!housel@cantor.gwydiondylan.org> attachments
@1085350027 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm also inclined to instead have people work on branches where reviewers could see it in situ before integrating to a main branch (or trunk), but I guess CVS isn't amenable to that. Perforce makes working that way rather easy.
@1085350110 <housel!housel@cantor.gwydiondylan.org> yes, that's the plan... small patches go in bz, larger work gets its own branch
@1085350128 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah.
@1085350155 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What's the cutoff point for "small"?
@1085350187 <housel!housel@cantor.gwydiondylan.org> that's open to interpretation
@1085350272 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Do you always speak in tautologies? :-)
@1085350337 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> btw, I'm working on dylan-mode, trying to figure out how to fix indenting of "local method" (it's broken in three or four ways) and that code is quite the twisty little maze.
@1085350349 <housel!housel@cantor.gwydiondylan.org> I can imagine
@1085350440 <housel!housel@cantor.gwydiondylan.org> For me currently the most annoying dylan-mode bug is '-matching
@1085350478 <housel!housel@cantor.gwydiondylan.org> apostrophes in comments or the header end up fontifying larget blocks of code as character constants
@1085350504 <housel!housel@cantor.gwydiondylan.org> s/larget/large/
@1085350538 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> As it happens, I was considering whether comments and headers should just be treated entirely as comments and suppress all interior fontification.
@1085350553 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What's typical in an emacs mode?
@1085350585 <housel!housel@cantor.gwydiondylan.org> no fontification in ocmments
@1085350629 <housel!housel@cantor.gwydiondylan.org> seeing "define method" fontified in comments is kind of consusing sometimes too
@1085350666 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, actually, I see that at least lisp-mode highlights binding names in comments. That still seems useful.
@1085350667 <housel!housel@cantor.gwydiondylan.org> I like being able to see at a glance when a block has been commented out, and fontified keywords makes one do a double-take
@1085350676 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> +1
@1085350776 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> btw, I'd kind of like more faces, to allow more differentiation. Currently, almost everything is given "function name" or "keyword" faces. Is it possible for modes to define mode-specific faces? (ie., that show up in Customize.)
@1085350827 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'll look into the apostrophe issue, since the indenting issue doesn't look like something I'll figure out soon.
@1085350929 <housel!housel@cantor.gwydiondylan.org> yes, modes can define more faces
@1085351002 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Okay, thanks, I'll look into that.
@1085351052 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I see the apostrophe bug if I put on in a "copyright:" header, but not if I put one in a /* block comment */.
@1085351058 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> s/on/one
@1085351073 <housel!housel@cantor.gwydiondylan.org> hmmm... how about // comments?
@1085351098 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Nope.
@1085351117 <housel!housel@cantor.gwydiondylan.org> hmmm... I might have misremembered, or it might be something else
@1085351122 <housel!housel@cantor.gwydiondylan.org> I'll keep an eye out
@1085351147 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought I'd seen it, too, but I can't repro it at this moment.
@1085351188 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Quotes " are a problem in headers, too.
@1085351213 <housel!housel@cantor.gwydiondylan.org> fortunately they are usually balanced there
@1085351312 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah. It should just not parse anything in a header value, though.
@1085351333 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What do you think of the idea of applying either the comment or document face to header values?
@1085351351 <housel!housel@cantor.gwydiondylan.org> what's the document face?
@1085351378 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In lisp-mode, it's used for top-level definition doc strings.
@1085351417 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> By default, the comment and document face have the same settings.
@1085351447 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...er, what I meant was: By default, the string and document faces have the same settings.
@1085351465 <housel!housel@cantor.gwydiondylan.org> so I see
@1085351469 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Where the string face is applied to string literals.
@1085351498 <housel!housel@cantor.gwydiondylan.org> doc-face seems fine to me
@1085351530 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The main thing for me is to try to use different faces where it makes sense, so users can decide whether to format them differently.
@1085351634 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> e.g., I set the string face to courier, dark gray, but the default font--for comments and code--is monaco, which I find much easier to read, generally.
@1085351690 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085351692 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This makes string literals stand out a bit and gives the characters more serifs, etc. to differentiate characters that may appear ambiguous in monaco.
@1085351740 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...and then I set the comment face to use the default font, but dark green, which isn't as noisome as the default bright red.
@1085351809 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh, that reminds me: lisp-mode uses the "warning" face (bold, bright red) for "error", "abort", et al. Do you think it makes sense to follow lisp-mode's lead here for dylan-mode?
@1085351865 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ie., it highlights calls to the error/abort functions.
@1085351964 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Say, this can't be right: dylan-mode checks "(> emacs-minor-version 30)" to decide whether or not to highlight in the header. Shouldn't it be comparing the major version as well?
@1085352522 <housel!housel@cantor.gwydiondylan.org> indeed it should
@1085352552 <housel!housel@cantor.gwydiondylan.org> who knows whether major-version was 18 or 19 when that was written
@1085352588 <housel!housel@cantor.gwydiondylan.org> yes, I like that idea (highlighting signal() and error())
@1085352604 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, I changed it, and now nothing is highlighted, including the keywords, yet the apostrophe still causes the body to be treated like the inside of a character literal. Feh.
@1085353162 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So, it turns out dylan-mode was intended to not highlight header contents (including keywords) on emacs 19.31 and later. I've fixed the test so it "works" now. However, I think I prefer having keywords highlighted. I'm wondering why they wanted everything off.
@1085359947 -zelazny.freenode.net- Riastradh (~riastradh@pool-141-154-214-206.bos.east.verizon.net) joined the channel
@1085364783 -zelazny.freenode.net- hannes (~hannes@port-195-158-171-122.dynamic.qsc.de) joined the channel
@1085375015 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION is quickly coming to the conclusion that emacs modes are futile to pursue
@1085375067 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> This stuff would be much simpler if the font-lock code worked on an AST instead of introducing all these ad-hoc, incomplete parsers.
@1085375787 <housel!housel@cantor.gwydiondylan.org> of course
@1085375824 <housel!housel@cantor.gwydiondylan.org> I told you about Semantic Bovinator before, right?
@1085375863 -dircproxy- You disconnected
@1085375987 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Sounds familiar; can you remind me?
@1085376058 -dircproxy- You connected
@1085376112 <housel!housel@cantor.gwydiondylan.org> http://cedet.sourceforge.net/semantic.shtml
@1085376311 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That looks...useful.
@1085377906 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> To somebody's credit, font-lock.el has a lot of useful commentary. I'm thinking the best approach may be to turn off font-lock's automatic fontification of strings and comments and add code to dylan-mode to handle them.
@1085378848 -zelazny.freenode.net- brucehoult (~chatzilla@metnet.met.co.nz) joined the channel
@1085379141 <housel!housel@cantor.gwydiondylan.org> hi bruce
@1085379412 <brucehoult!~chatzilla@metnet.met.co.nz> hi Peter
@1085379428 <brucehoult!~chatzilla@metnet.met.co.nz> speaking at you from the new job
@1085379453 <brucehoult!~chatzilla@metnet.met.co.nz> ... mindystrap proceeding...
@1085379643 <brucehoult!~chatzilla@metnet.met.co.nz> .. and going home :-)
@1085380993 -zelazny.freenode.net- bhoult (~bruce@202-0-60-58.adsl.paradise.net.nz) joined the channel
@1085387090 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yay! I figured out how to suppress fontification of interchange file headers. This prevents apostrophes and other things from "leaking" into the body.
@1085387214 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The trick was to figure out which font-lock default function to override in order to restrict it to only fontifying the body of the file.
@1085387253 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And then there's a global variable you have to set to prevent it from ignoring you.
@1085387259 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Whee!
@1085391476 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1085393177 <^self!~fn@66.209.66.105> bot bot bot.
@1085393184 <^self!~fn@66.209.66.105> i need lunch.
@1085394569 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Evening!
@1085395152 [cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net] ACTION is off to bed after several hours of battling it out with emacs
@1085396330 -zelazny.freenode.net- bhoult (~bruce@202-0-61-1.adsl.paradise.net.nz) joined the channel
@1085396354 -zelazny.freenode.net- bhoult (~bruce@202-0-61-1.adsl.paradise.net.nz) joined the channel
@1085396530 -zelazny.freenode.net- bhoult (~bruce@202-0-61-1.adsl.paradise.net.nz) left the channel
@1085396667 -zelazny.freenode.net- bhoult (~bruce@202-0-61-1.adsl.paradise.net.nz) joined the channel
@1085396691 <bhoult!~bruce@202-0-61-1.adsl.paradise.net.nz> so Chris's recent change to melange breaks mindystrapping because of the #e2
@1085396705 <bhoult!~bruce@202-0-61-1.adsl.paradise.net.nz> what's the best approach to fix it?
@1085396712 <bhoult!~bruce@202-0-61-1.adsl.paradise.net.nz> conditional code in melange?
@1085396816 <bhoult!~bruce@202-0-61-1.adsl.paradise.net.nz> extend the Mindy lexer to accept bignum syntax for literals but just make a regular integer?
@1085396899 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085400892 -zelazny.freenode.net- Riastradh (~riastradh@pool-151-203-222-80.bos.east.verizon.net) joined the channel
@1085401259 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1085401267 <^self!~fn@66.209.66.105> hi
@1085401274 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> g'day
@1085401428 <^self!~fn@66.209.66.105> the output of python Mnet, you mean?
@1085401434 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> cpage: besides fontifying signal and error, it would be cool to highlight the use of exit variables in 'block's.  e.g., block(return) ... return(foo) end
@1085406730 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085409368 <^self!~fn@66.209.66.105> ok.
@1085409832 -dircproxy- You disconnected
@1085410994 -dircproxy- You connected
@1085414833 -dircproxy- You disconnected
@1085416819 -dircproxy- You connected
@1085418232 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1085420704 -zelazny.freenode.net- hannes (hannes@cantor.gwydiondylan.org) joined the channel
@1085420708 <hannes!hannes@cantor.gwydiondylan.org> hi
@1085420804 <housel!housel@cantor.gwydiondylan.org> hi hannes
@1085424405 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> bhoult: I guess either approach is fine for getting Mindy to handle Melange. For the short term I guess just use conditional code. It doesn't seem that the case that code handles shows up in a bootstrap, and rarely elsewhere.
@1085424491 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> cgay: Do you mean specifically that you want block exit functions to be given font-lock-warning-face like "error"?
@1085424513 <cgay!~cgay@ita4fw1.itasoftware.com> calls to them, yes
@1085424545 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Would it suffice to just highlight calls to any function named "return"?
@1085424565 <cgay!~cgay@ita4fw1.itasoftware.com> in java mode i turned off all colorizing except new bindings and non-local exits and it worked quite nicely.
@1085424624 <cgay!~cgay@ita4fw1.itasoftware.com> i often name the exit function "break" if i'm using it in such a way that it doesn't exit the entire method
@1085424653 <cgay!~cgay@ita4fw1.itasoftware.com> maybe break, return, exit, and continue, if that's a lot easier.  but i've seen them named exit-inner, exit-outer, etc
@1085424823 <^self!~fn@66.209.66.105> hmm.  
@1085424838 <^self!~fn@66.209.66.105> is andreas porting fun-o's networking library to gd?
@1085424853 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually parsing and keeping track of binding names would be a bit harder. AFAICT, none of the modes really do that sort of thing during fontification. Although, dylan-mode looks for begin words in order to fill in "end ..." while editing.
@1085424937 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In fact, lisp-mode and c-mode mostly only highlight things at point of definition, not usage.
@1085424982 <cgay!~cgay@ita4fw1.itasoftware.com> i guess that's the only reason i've ever seen that having everything be a reserved word is a win ;-)
@1085425005 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Well, yes, having an underpowered language makes it easier to write cheap parsers for them. :-)
@1085425024 <cgay!~cgay@ita4fw1.itasoftware.com> yeah
@1085425051 <cgay!~cgay@ita4fw1.itasoftware.com> or having a prefix syntax
@1085425058 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Actually, c-mode is quite complicated, not necessarily due to the language power, but due to the irregularity and complexity of the syntax, which is really the deciding factor.
@1085425105 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> AFAICT there's not a huge difference in the complexity of lisp-mode vs. dylan-mode so far. Lisp is deceptively complex.
@1085425166 <cgay!~cgay@ita4fw1.itasoftware.com> you mean deceptively simple and actually complex
@1085425179 <Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net> If you just used an editor that edited S-expressions instead of text, a lisp-mode would be very easy...
@1085425179 <^self!~fn@66.209.66.105> i see.
@1085425180 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It is complex, and deceptive about it. :-)
@1085425224 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> If you just used an editor that edited ASTs instead of text, most modes would be much easier.
@1085425271 <cgay!~cgay@ita4fw1.itasoftware.com> i think ^self is having a conversation with ^himself
@1085425293 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "S-expressions" is another way of saying "some parsed form where semantic parsing devolves into looking up the definition of each already-parsed lexeme."
@1085425369 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The problem with text-based modes is that there is no full compiler around with definition tables, so you have to do the lexing and parsing over and over.
@1085425411 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> At some point I'm going to see what Semantic Bovinator offers as an alternative and how much work it would take to add Dylan support.
@1085425808 <housel!housel@cantor.gwydiondylan.org> ^self: the FD network library is partially ported to d2c in the newio branch, but it's not finished yet and it wouldn't be very useful without threads
@1085427406 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085427481 <^self!~fn@66.209.66.105> yeah, i was wondering about threads.
@1085427994 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085428175 -zelazny.freenode.net- CIA-7 (micah@hyperreal.info) joined the channel
@1085428987 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085429736 -zelazny.freenode.net- CIA-6 (micah@hyperreal.info) joined the channel
@1085430451 -zelazny.freenode.net- CIA-6 (micah@hyperreal.info) joined the channel
@1085430814 -zelazny.freenode.net- CIA-6 (micah@hyperreal.info) joined the channel
@1085432559 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085435296 <brucehoult!~chatzilla@metnet.met.co.nz> cpage: so in the end I replaced your #e2 with as(<extended-integer>, 2)
@1085435313 <brucehoult!~chatzilla@metnet.met.co.nz> Which will be a tad slower, but as you say it's a rare case anyway
@1085435353 <brucehoult!~chatzilla@metnet.met.co.nz> It's very strange that Mindy has <extended-integer> but doesn't have either a syntax for literals or a way to represent them in the bytecode.
@1085436675 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> bruceholt: Yeah, I saw that. That's fine.
@1085436848 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> BTW, are <extended-integer> just a vector of integers?
@1085437019 <housel!housel@cantor.gwydiondylan.org> in d2c, a vector of 16-bit (short) integers
@1085437053 <housel!housel@cantor.gwydiondylan.org> (thanks to Bruce; before, it was 8-bit bytes)
@1085437101 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 16-bit, eh? What's the advantage over, say, native machine words?
@1085437124 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (Or just whatever a non-extended integer uses.)
@1085437332 <housel!housel@cantor.gwydiondylan.org> 32 x 32 = 64 bit result
@1085437351 <housel!housel@cantor.gwydiondylan.org> not guaranteed representable in C
@1085437381 <housel!housel@cantor.gwydiondylan.org> also 32 + 32 = 33
@1085437414 <housel!housel@cantor.gwydiondylan.org> relatively inefficient to determine using C
@1085437471 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah. "32 + 32 = 33"?
@1085437521 <housel!housel@cantor.gwydiondylan.org> carry bit on addition/subtraction
@1085437710 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah.
@1085437774 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It's kind of surprising that C doesn't readily support implementing math at that level. There's no explicit way to get the carry bit.
@1085437983 <housel!housel@cantor.gwydiondylan.org> c-- has that capability
@1085438038 <housel!housel@cantor.gwydiondylan.org> I wouldn't say it's all that suprising, it would require a builtin function like ldiv(), or multiple values, or some similar solution, and very few people need it
@1085438241 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I'm surprised because C is such a low-level language in other respects. I'm looking forward to seeing C-- used everywhere.
@1085438319 -zelazny.freenode.net- pritlove_ (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085438411 <brucehoult!~chatzilla@metnet.met.co.nz> yeah, I first extended it from 8 to 15 bits
@1085438420 <brucehoult!~chatzilla@metnet.met.co.nz> .. and then realised that 16 was no problem
@1085438426 <brucehoult!~chatzilla@metnet.met.co.nz> oh!
@1085438462 <brucehoult!~chatzilla@metnet.met.co.nz> I remember fixing a bug in the division code ... and last night I saw similar-looking code in Mindy interp.  I wonnder if it has the same bug?
@1085438506 <cgay!~cgay@ita4fw1.itasoftware.com> pardon my ignorance, but what is C--?
@1085438516 <housel!housel@cantor.gwydiondylan.org> http://www.cminusminus.org/
@1085438550 <housel!housel@cantor.gwydiondylan.org> it's a language (and a compiler, QC--) intended to replace C as a "high-level assembly language" target of code generators
@1085438581 <cgay!~cgay@ita4fw1.itasoftware.com> cool
@1085439013 <cgay!~cgay@ita4fw1.itasoftware.com> peter, did you have a chance to work on the Windows release?
@1085439043 <housel!housel@cantor.gwydiondylan.org> I did, but I encountered bug 7024, and spent a lot of time trying to fix it so I didn't quite finish
@1085439054 <housel!housel@cantor.gwydiondylan.org> I'll try again this evening
@1085439155 <cgay!~cgay@ita4fw1.itasoftware.com> 7024, 7024... ah, ok
@1085443828 -zelazny.freenode.net- cpage_ (~Chris@charger.intuit.com) joined the channel
@1085446870 <cpage_!~Chris@charger.intuit.com> I can't believe I'm asking this, but someone asked me and I realized I don't know definitively: Does d2c do tail-call optimization?
@1085447553 -zelazny.freenode.net- cpage__ (~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net) joined the channel
@1085448113 <housel!housel@cantor.gwydiondylan.org> yes, within a single function or a group of local functions
@1085448485 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Great, thanks.
@1085450657 -dircproxy- You disconnected
@1085457838 -zelazny.freenode.net- CIA-7 (micah@hyperreal.info) joined the channel
@1085458450 -dircproxy- You connected
@1085466271 <^self!~fn@66.209.66.105> hmm.
@1085466273 <^self!~fn@66.209.66.105> From: Davinator (dpassage@63.90.156.234) 
@1085466275 <^self!~fn@66.209.66.105> Air New Zealand has bought ad signs in the BART stations in San Francisco that say "Defectors Welcome".
@1085467776 -zelazny.freenode.net- bhoult (~bruce@202-0-60-173.adsl.paradise.net.nz) joined the channel
@1085468057 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ouch!
@1085468060 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I like it :-)
@1085468346 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> hmm?
@1085468423 -dircproxy- You disconnected
@1085468493 [bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz] ACTION just checked the log
@1085468525 [bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz] ACTION not sure that we want NZ to be californicated
@1085468535 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> heh
@1085468566 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> those from AZ/NM/CO/NV/OR/WA definitely welcome :-)
@1085468573 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> BC, too...
@1085468605 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> You're just saying that because there's less of them. :-)
@1085468621 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> well, each of those states is about the same population as NZ
@1085468651 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> well AZ and OR anyway.  WA is quite a bit bigger
@1085468657 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> And probably less than the population of CA in total. (Just a WAG.)
@1085468684 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> CA folk ... well a lot of them I'd welcome...   If I can give them a quick quiz first
@1085468834 -dircproxy- You connected
@1085468849 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> e.g.?
@1085468905 <housel!housel@cantor.gwydiondylan.org> habla engles?
@1085469033 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> e.g., what would be on the quiz?
@1085469208 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I'd have to think about the exact questions, but they might concern desired tax rates, tendency to attempt to get laws passed to make people do what's good for them, attitude towards gun control.  That sort of thing.
@1085469286 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> oh yeah: "so why do you want to move to NZ rather than to AZ/NM/CO/NV/OR/WA?" migth be one of them :-)
@1085469641 <^self!~fn@66.209.66.105> is "lord of the rings" a good answer for that question?
@1085469649 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Shouldn't the answer to the last question be obvious?
@1085469667 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> It's interesting to compare gwydion running on my new dual 2.0 GHz G5 at work and my new 2.2 GHz Athlon XP at home
@1085469669 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Because it's NZ."
@1085469678 <andreas!andreas@cantor.gwydiondylan.org> Morning!
@1085469687 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> morning Andreas!
@1085469697 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> How do they compare?
@1085469730 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I don't have the G5 mindystrap time at the moment, but for doing d2c stuff the Athlon is twice as fast
@1085469759 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> regular build of d2c: Athon 6:05, G5 12:30
@1085469784 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> build hello-world: Athlon 3.3 sec, G5 6 sec
@1085469809 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the Athlon does a mindystrap in 1:56
@1085469844 <^self!~fn@66.209.66.105> that's...not what i would've expected.
@1085469849 <^self!~fn@66.209.66.105> is os x really that bad?
@1085469880 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> that makes the G5 pretty comparable to a 3 or 3.2 GHz P4
@1085469904 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> a *little* slower, perhaps.
@1085469931 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I tried a d2c regular build on Chris Double's P4 3.2 and it took 10:30
@1085469947 <^self!~fn@66.209.66.105> i guess i expected the g5 to trounce the athlon.
@1085469958 <housel!housel@cantor.gwydiondylan.org> I went ahead and updated http://www.gwydiondylan.org/~housel/fundev-snapshot.html for the release I did this evening, but the upload won't be done for another hour or so
@1085469969 <housel!housel@cantor.gwydiondylan.org> and it's way past my bedtime
@1085469984 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Are the hard drives, RAM, and busses comparable? Did you compare VM swapping activity?
@1085469994 <andreas!andreas@cantor.gwydiondylan.org> Hi Peter! I'll test drive it today.
@1085470012 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> no VM activity and the G5 has twice the RAM anyway (1.5 GB)
@1085470023 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> both have 160 GB disks
@1085470027 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> No VM activity during a bootstrap?
@1085470037 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> not even close to VM activity
@1085470042 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What about the RPM and buss of the hard drives?
@1085470043 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> hundreds of MB free
@1085470075 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> both AT100 I think
@1085470082 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> it's irrelevant anyway
@1085470096 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> hard disk speed isn't afactor for running d2c
@1085470117 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> That's surprising.
@1085470162 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the Athlon has a 200 MHz FSB (advertised as 400).  The G5 has a 1 GHz FSB
@1085470171 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> both have 512 KB L2 cache
@1085470211 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> if you want to look at RAM streaming speeds, the Athlon does 3.2 GB/sec max
@1085470219 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the G5 is .. uh ... 6 or 8 GB/sec
@1085470229 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> so is a P4, btw
@1085470261 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> here's what I suspect the difference is:
@1085470275 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the Athlon has 10 pipe stages, the G5 and P4 20+
@1085470318 <andreas!andreas@cantor.gwydiondylan.org> Which makes branches much cheaper on the Athlon.
@1085470325 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> if we could get rid of a bunch of branches from d2c code, or make them better predicted then I think it would be a lot more even
@1085470338 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I can only assume prediction would be a big factor.
@1085470345 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> yep
@1085470351 <andreas!andreas@cantor.gwydiondylan.org> It's hard to get the branch prediction across the C code. :)
@1085470352 <housel!housel@cantor.gwydiondylan.org> gcc 3.4 can do profiling-based feedback
@1085470357 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The PowerPC can do some unnecessary speculation without a big speed hit.
@1085470370 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> our code has a ton of branches, especuially for error checking
@1085470372 <housel!housel@cantor.gwydiondylan.org> haven't actually tried it yet
@1085470386 <andreas!andreas@cantor.gwydiondylan.org> PPC has a bit to indicate whether branches are expected to be taken or not.
@1085470404 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I suspect that if we took all those inlined tests and put them in functions it would speed up all three current architectures
@1085470411 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> ... but especially the P4 and G5
@1085470418 [housel!housel@cantor.gwydiondylan.org] ACTION goes to bed... upload ETA 00:47:05
@1085470426 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I plan to try it anyway
@1085470430 <andreas!andreas@cantor.gwydiondylan.org> Night Peter!
@1085470439 <hannes!hannes@cantor.gwydiondylan.org> morning
@1085470444 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Why would removing inlining improve the branching situation?
@1085470455 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I thought inlining removed branching overhead?
@1085470467 <andreas!andreas@cantor.gwydiondylan.org> Less load on the 1st level cache?
@1085470467 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> because there would then be far fewer distinct branches executed
@1085470487 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> which means all those error-checking tests would be correctly predicted
@1085470495 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> (as no error)
@1085470514 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I don't follow. That's okay, though. I believe you.
@1085470517 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> and each CPU has a branch prediction cache that can hold maybe 512 different branches
@1085470540 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> it's not *branches* that is the problem ... it's CONDITIONAL branches
@1085470561 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> unconditional ones such as subroutine call/return are fine
@1085470577 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It's not *conditional* branches...it's MIS-PREDICTED conditional branches ;-)
@1085470586 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> true
@1085470601 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I see your point, but still find it surprising.
@1085470614 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the G5 has some extra static branch prediction bits btw
@1085470637 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In a register, you mean?
@1085470641 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> you can say "predict this branch as taken -- and dont' even bother to put it in the branch prediction tables"
@1085470658 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> no: two bits in the opcode that were previously undefined
@1085470664 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Oh.
@1085470676 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I guess I better go read up on the new CPU.
@1085470708 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> if we could get the C code to set those bits for the error tests that would be the best
@1085470714 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Does that mean that you can say "predict as taken -- but if you subsequently discover it's usually not, ignore this flag"?
@1085470732 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> that's what the *old* hints bits did
@1085470746 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "bits" or "bit"?
@1085470747 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the ones first on the 601
@1085470776 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> well, you need two
@1085470807 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> no, wait, that was just one bit
@1085470826 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the PPC601 by default predicted backward branches as taken, forward ones as not taken
@1085470841 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the bit in the instruction could say "reverse your prediction"
@1085470853 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right.
@1085470891 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the 603/604/750/7400 just built on that, adding caches and so forth
@1085470904 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Yeah, that was what I was getting at.
@1085470913 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> if there was history in the cache they used it, otherwise they did a static prediciton like the 601
@1085470939 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> but the 970 has entirely new bits added to that
@1085470942 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> So the G5 has one new bit to say "only pay attention to the flag bit"?
@1085470972 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> not sure
@1085470982 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I thought I read it was two new ones, but maybe it's only two total
@1085471002 [bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz] ACTION never had access to a 970 until yesterday, so didn't care
@1085471044 [bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz] ACTION wonders whether gcc has a directive
@1085471050 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> brb
@1085471068 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The Mac OS X developer tools includes some really nice performance tools.
@1085471087 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think there's a way to see if there are lots of mis-predictions.
@1085471438 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> yeah, I'll put CHUD onto it shortly
@1085471487 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> btw, my reply to people who answer "Lord of the Rings" would be to make sure they've seen _Meet the Feebles_ as well
@1085471524 <andreas!andreas@cantor.gwydiondylan.org> It's on my list.
@1085471536 <^self!~fn@66.209.66.105> heh.
@1085471543 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> We should contribute to this: http://pleac.sourceforge.net/
@1085471604 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> did you know I knew Nathan Torkington before he was famous?
@1085471608 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> he's a VUW grad
@1085473411 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> From a comparison of my older Motorola PowerPC manual and the 970 manual, the older CPUs have a single branch-prediction bit, which inverts the default behavior of the CPU, as we mentioned...
@1085473436 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...the 970 adds one new bit, the "a" bit, which is used in concert with the existing bit "t" to produce four values:
@1085473452 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 00 - No hint is given
@1085473456 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 01 - Reserved
@1085473463 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 10 - The branch is very likely not to be taken
@1085473472 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 11 - The branch is very likely to be taken
@1085473536 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I dont' see any way to make that encoding play nicely with the older encoding
@1085473547 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> From a quick reading it seems that existing code will have either 00 or 01, making them effectively ignored on the 970.
@1085473582 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> 00 is compatible in the sense that the old 0 meant "no hint given".
@1085473589 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> which means that 01 can not in fact be reserved ... it's used in lots of existing code
@1085473601 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I presume they mean they intend to repurpose it in the future.
@1085473610 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> .. and means "reverse your default static hint"
@1085473643 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Right, so I think what this means is the 970 ignores those old hints.
@1085473668 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Because they probably weren't as useful as the branch prediction cache.
@1085473691 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> ...or aren't, on the 970.
@1085473802 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Ah, there's a Compatibility Note a couple pages later.
@1085473877 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> From the note:
@1085473880 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> "Because these bits have always been defined either to be ignored or to be treated as hints, a given program will produce the same result on any implementation regardless of the values of the bits. Also, because even the ? y ? bit is ignored, in practice, by most processors that implement versions of the architecture that precede Version 2.00, the performance of a given program on those processors will not be affected by the va
@1085473880 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> lues of the bits."
@1085474022 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think what they're saying here is "y was noise, but now we've added another bit you can use to really provide hints" ;-)
@1085474121 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> except the new one isn't a hint
@1085474155 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> It's a mask for the existing hint bit. :-)
@1085474165 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> if you set it to 10 or 11 then the 970 *will* follow the branch you tell it to
@1085474183 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> What it really does is give you a tristate value instead of the .5 value you had before.
@1085474191 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> right
@1085474251 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> In the 980 those bits will be ignored, too, and a bit in the previous instruction will give you 8 values. :-)
@1085474402 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> The 970 manual doesn't mention whether it actually ignores the prediction cache, though.
@1085474409 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Where'd you read that?
@1085474420 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> (that I can see)
@1085474464 <^self!~fn@66.209.66.105> i'm just hoping for a g5 to fall out of the back of a truck or something.
@1085474600 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I paid for the Select program, which gets me a hardware discount (~20%). I'm trying to decide whether to wait until after WWDC to see what new comes along, or buy one now. I've put off buying a new machine for a couple years and my 500MHz G4 is really starting to feel creaky as I work on newer, bigger projects.
@1085474676 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> I really love OSX, but I really love the cheep speed of a self-built x86 machine
@1085474725 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> the new Athlon box cost me about US$1k
@1085474750 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> and, for d2c right now, it's twice the speed of a $3k box from Apple
@1085474812 <bhoult!~bruce@202-0-60-173.adsl.paradise.net.nz> for everything *except* Dylan development, my 266 MHz PowerBook runs OSX just fine
@1085476871 [^self!~fn@66.209.66.105] ACTION shakes fist at linux's traffic shaping
@1085486537 -zelazny.freenode.net- hannes (hannes@cantor.gwydiondylan.org) joined the channel
@1085492298 -zelazny.freenode.net- lisppaste (~lisppaste@lamancha.opendarwin.org) joined the channel
@1085492840 -zelazny.freenode.net- chandler (~chandler@chandler.registered) joined the channel
@1085493943 <^self!~fn@66.209.66.105> home, must feed.
@1085494978 -dircproxy- You disconnected
@1085497774 <cgay!~cgay@ita4fw1.itasoftware.com> housel: thanks for putting up the new windows release
@1085497838 <cgay!~cgay@ita4fw1.itasoftware.com> one minor problem that was easy to work around: the zipped folder was "Functional Developer" instead of "Functional Objects".
@1085497922 <cgay!~cgay@ita4fw1.itasoftware.com> also, i guess fun-o used to distribute the gnu linker with FD but now we don't
@1085499285 <cgay!~cgay@ita4fw1.itasoftware.com> looks like one just needs to install binutils.  i'll give that a whack later
@1085501227 -zelazny.freenode.net- CIA-8 (micah@hyperreal.info) joined the channel
@1085503633 -dircproxy- You connected
@1085503683 <housel!housel@cantor.gwydiondylan.org> actually, I haven't had much luck getting it to work with the gnu linker
@1085503697 <housel!housel@cantor.gwydiondylan.org> I'm sure I'm doing something wrong in my build, but I'm not sure what
@1085503972 <housel!housel@cantor.gwydiondylan.org> also, the zip file is intended to be unzipped within C:\Program Files\Functional Objects
@1085504018 <housel!housel@cantor.gwydiondylan.org> hmmm... I guess I'd better change the web page then
@1085504487 <housel!housel@cantor.gwydiondylan.org> done.
@1085504719 <cgay!~cgay@ita4fw1.itasoftware.com> this suggests one can't get the MS linker without VC++ anymore... http://www.ai.mit.edu/~gregs/info-dylan-archive-html-2001/msg00547.html
@1085504740 <cgay!~cgay@ita4fw1.itasoftware.com> i wonder if i should even try installing the latest Core SDK
@1085504876 <housel!housel@cantor.gwydiondylan.org> The latest core SDK (the Feb. 2003 edition) comes with a 64-bit linker and C compiler
@1085504887 <housel!housel@cantor.gwydiondylan.org> I don't know if they work with 32-bit targets or not
@1085504960 <housel!housel@cantor.gwydiondylan.org> duh... I should have read that article first
@1085505022 <housel!housel@cantor.gwydiondylan.org> I suspect it may be subject to the same problem that the VC7 linker is, namely it will only produce separate debugging info in .PDB files, and we're not properly copying the .PDB
@1085505056 <housel!housel@cantor.gwydiondylan.org> I plan on fixing all of these problems when/as I finish the new build system
@1085505862 <housel!housel@cantor.gwydiondylan.org> lovely... Linda Scheimann is continuing to serve as comp.lang.dylan's "resident kook"
@1085505934 -zelazny.freenode.net- sprout (~Snak@drop242.drizzle.com) joined the channel
@1085506938 <cgay!~cgay@ita4fw1.itasoftware.com> sort of seems like she's replying to some other thread.
@1085507000 <cgay!~cgay@ita4fw1.itasoftware.com> it must be too deep for me
@1085507032 <housel!housel@cantor.gwydiondylan.org> no, she's completely off in left field
@1085507047 <housel!housel@cantor.gwydiondylan.org> as far as I can tell
@1085507062 <housel!housel@cantor.gwydiondylan.org> all of her postings are like that
@1085507086 <cgay!~cgay@ita4fw1.itasoftware.com> at least she's not too Naggumesque
@1085507459 -zelazny.freenode.net- pritlove (~pritlove@62.80.42.113) joined the channel
@1085511554 -zelazny.freenode.net- pritlove (~pritlove@dsl-213-023-157-111.arcor-ip.net) joined the channel
@1085512195 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> I think Linda's a bot.
@1085512211 <cpage!~Chris@adsl-67-118-119-204.dsl.snfc21.pacbell.net> Where else does she post?
@1085512623 <cgay!~cgay@ita4fw1.itasoftware.com> just being able to use the IDE, without linking, is still quite useful :)
@1085515937 -zelazny.freenode.net- CIA-7 (micah@hyperreal.info) joined the channel
@1085516146 -zelazny.freenode.net- CIA-2 (micah@hyperreal.info) joined the channel
@1085516154 <housel!housel@cantor.gwydiondylan.org> hmmm... looks like she's all over the map, but heavily into "microsoft.test"
@1085516347 <housel!housel@cantor.gwydiondylan.org> heh... she has also been known to post to "alt.usenet.kooks"
@1085516363 <housel!housel@cantor.gwydiondylan.org> but enough of that...
@1085518733 -zelazny.freenode.net- oudeis (~oudeis@bzq-218-95-152.red.bezeqint.net) joined the channel
@1085518773 <housel!housel@cantor.gwydiondylan.org> hi
@1085519255 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> hello
@1085520247 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: i've installed the hacker release (20040524). built mps. now i try to build-release and can't make it do it without cvs access
@1085522219 <housel!housel@cantor.gwydiondylan.org> re
@1085522236 <housel!housel@cantor.gwydiondylan.org> what's it trying to check out?
@1085522354 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> library-packs
@1085522438 <housel!housel@cantor.gwydiondylan.org> That should be in Sources\Library-Packs; is it not?
@1085522486 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> it is. maybe i specified sources incorrectly
@1085522499 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> can you give me the incantation you use?
@1085522511 <housel!housel@cantor.gwydiondylan.org> cd admin\builds
@1085522528 <housel!housel@cantor.gwydiondylan.org> build-release.bat c:\fundev /generations 1 /exports /target internal-release
@1085522613 <housel!housel@cantor.gwydiondylan.org> make sure there are no spaces in the path to the source dir; I use "c:\fundev" for that reason
@1085522632 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> well it keeps complaining i don't have cvs installed on my path (i don't)
@1085522656 <cgay!~cgay@ita4fw1.itasoftware.com> i think there's an option (or env var) for turning off cvs checks
@1085522692 <housel!housel@cantor.gwydiondylan.org> I haven't tried it with cvs removed from my path yet
@1085522823 <cgay!~cgay@ita4fw1.itasoftware.com> hmm. looks like i was mis-remembering
@1085522864 -zelazny.freenode.net- pritlove (~pritlove@62.206.114.179) joined the channel
@1085522911 <housel!housel@cantor.gwydiondylan.org> those batch files are annoyingly complex
@1085522946 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok. i'll just try to install cvs on the path then 
@1085522946 <cgay!~cgay@ita4fw1.itasoftware.com> yeah
@1085522964 <cgay!~cgay@ita4fw1.itasoftware.com> you could try hacking cvs-checkout.bat with "goto :end" :)
@1085522979 <housel!housel@cantor.gwydiondylan.org> fdmake.pl is a lot simpler, but it doesn't yet account very well for all of the various tools that need to be run (parser-compiler, console-scepter)
@1085523015 <housel!housel@cantor.gwydiondylan.org> or for the stuff that happens at the end of a release build (preparing Examples and Redistributable directories etc.)
@1085523128 <cgay!~cgay@ita4fw1.itasoftware.com> we should be able to write this stuff in dylan
@1085523133 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> could't all this be written in dylan may i ask?
@1085523147 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> :):):)
@1085523192 <housel!housel@cantor.gwydiondylan.org> of course it could, the compiler's project manager can already do most of it, but it might well run out of memory in the process
@1085523295 <housel!housel@cantor.gwydiondylan.org> that's why I wrote fdmake.pl; otherwise you could set the environment-variables and do "console-compiler -build console-compiler"
@1085523426 <housel!housel@cantor.gwydiondylan.org> do programs compiled with FunDev and MPS always give 0.5GB coredumps on linux?
@1085523435 <housel!housel@cantor.gwydiondylan.org> they do in the linuxulator
@1085523497 <housel!housel@cantor.gwydiondylan.org> oudeis: it's also worth noting that the release contains some uncommitted patches
@1085523532 <hannes!hannes@cantor.gwydiondylan.org> housel: hi, did you try to run a program with threads on linuxulator with mps-fd?
@1085523558 <housel!housel@cantor.gwydiondylan.org> hannes: I tried gctest but it didn't work very well
@1085523792 <housel!housel@cantor.gwydiondylan.org> oudeis: Functional Developer\Sources in the release zipfile contains the sources used to build the release, though
@1085523861 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: i've installed cvs, it is churning on now
@1085524322 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> got error linking dxdylan.dll
@1085524336 <housel!housel@cantor.gwydiondylan.org> what does the log file say?
@1085524359 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> lots of things
@1085524382 <housel!housel@cantor.gwydiondylan.org> at the end?
@1085524387 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> can't connect to lisppaste to paste it
@1085524423 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> it's kinda hard for me to paste it now - my irc is on unix box
@1085524431 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> just a moment
@1085524589 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i've put the log at oudeis.maclisp.org/compile-dylan.log
@1085524726 <housel!housel@cantor.gwydiondylan.org> got it
@1085524763 <housel!housel@cantor.gwydiondylan.org> is "link" in your path?  what version does it say it is?
@1085524815 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> 7.10.3077
@1085524871 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> this is visual studio 7 - may this be the problem? as far as i understand, you use vs6
@1085524888 <housel!housel@cantor.gwydiondylan.org> I use VS6 to build the release, yes
@1085524931 <housel!housel@cantor.gwydiondylan.org> As the web page explains, VS7 doesn't work for building releases
@1085524989 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> yes, soory, now i see it
@1085524994 <housel!housel@cantor.gwydiondylan.org> I'm able to build programs in the IDE with VS7, but not releases
@1085525024 <housel!housel@cantor.gwydiondylan.org> also programs built in the IDE with the VS7 linker don't have debugging info so you can't single-step them
@1085525102 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok, i'll try to find vs6 installation. maybe this will happen only day after tomorrow when i get to work :(
@1085525140 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> btw does anyone work on nice windows installer for the snapshot release?
@1085525205 <housel!housel@cantor.gwydiondylan.org> not yet, that I know of
@1085525216 <housel!housel@cantor.gwydiondylan.org> InstallShield costs $1000 or so
@1085525222 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i think i'll look into it in the meantime
@1085525231 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> microsoft installer is free
@1085525244 <housel!housel@cantor.gwydiondylan.org> microsoft installer is just a bunch of APIs
@1085525252 <housel!housel@cantor.gwydiondylan.org> we could come up with bindings for them, of course
@1085525258 <housel!housel@cantor.gwydiondylan.org> that might be a good way to go
@1085525272 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> nullsoft install system is simple, but works
@1085525282 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> the one used for winamp
@1085525321 <housel!housel@cantor.gwydiondylan.org> URL?
@1085525335 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> just google for it
@1085525377 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> nsis.sourceforge.net/home
@1085525452 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> shoutcast, divx. city of heroes are listed as users
@1085525469 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> must be pretty sound stuff
@1085525563 <housel!housel@cantor.gwydiondylan.org> nifty
@1085525623 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> do you want me to try and build a simple script - just implementing your installation  instructions + file assics for hdp and dylan?
@1085525639 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> s/assics/associations
@1085525678 <housel!housel@cantor.gwydiondylan.org> sure
@1085525693 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085525722 <housel!housel@cantor.gwydiondylan.org> (I keep meaning to copy the file associations from FD2.0 into my fd.reg but never got around to it)
@1085525766 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> there's small bug in this release - it doesn't find standalone-deuce example, even though it is there
@1085525792 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i mean if you try to open it from the startup dialog
@1085525817 <housel!housel@cantor.gwydiondylan.org> yes, I'd noticed that myself
@1085525835 <housel!housel@cantor.gwydiondylan.org> I'll submit a bug entry
@1085525933 <housel!housel@cantor.gwydiondylan.org> browsing to find the source and building it works, though
@1085525973 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> yes
@1085526139 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> no, sorry - it can't find gnu installation, presumably for linking, even though i have cygwin installed
@1085526151 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> where do i give it the path to cygwin?
@1085526357 <housel!housel@cantor.gwydiondylan.org> You could copy the "tools" directory from FD2.0SP3
@1085526368 <housel!housel@cantor.gwydiondylan.org> that's one of the places it looks for it
@1085526384 <housel!housel@cantor.gwydiondylan.org> It wouldn't work at all with a modern cygwin
@1085526398 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085526415 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> why didn't you include it in the snapshot?
@1085526444 <housel!housel@cantor.gwydiondylan.org> because it didn't work for me, even when I did that
@1085526449 <housel!housel@cantor.gwydiondylan.org> otherwise I would have
@1085526476 <housel!housel@cantor.gwydiondylan.org> I'll have to do a bit more experimentation
@1085526506 <housel!housel@cantor.gwydiondylan.org> ok, it's bug 7029
@1085526511 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i don't understand. what linker do you use then? msvc6?
@1085526533 <housel!housel@cantor.gwydiondylan.org> yes
@1085526557 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> and when i install it, what should i do to make fundev use it?
@1085526574 <housel!housel@cantor.gwydiondylan.org> just make sure it's in your path
@1085526578 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085526585 <housel!housel@cantor.gwydiondylan.org> to build a release, you need the System SDK too
@1085526607 <housel!housel@cantor.gwydiondylan.org> and its directories should be included in the VC search path directories INCLUDE and LIB
@1085526653 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i have converted all my projects (10+, work) from vs6 to vs7 and thought i'll never use it again :-)
@1085526684 <housel!housel@cantor.gwydiondylan.org> I bought a copy of VS6 on eBay expressly for this purpose
@1085526838 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> which version of system sdk do i need? do you have the url?
@1085527001 <housel!housel@cantor.gwydiondylan.org> February 2003 version
@1085527004 <housel!housel@cantor.gwydiondylan.org> http://www.microsoft.com/msdownload/platformsdk/sdkupdate/default.htm
@1085527034 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085528842 -dircproxy- You disconnected
@1085530067 -dircproxy- You connected
@1085530930 <brucehoult!~chatzilla@metnet.met.co.nz> G5 mindystrap takes 2h26
@1085530950 <brucehoult!~chatzilla@metnet.met.co.nz> on my XP3200+ it's 1h56
@1085530966 <brucehoult!~chatzilla@metnet.met.co.nz> so they're much closer on that than they are for d2c builds
@1085531306 <housel!housel@cantor.gwydiondylan.org> it would be interesting to try out gcc's new profile-directed optimization feedback to see how much of a difference that makes
@1085532358 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: what license should the installer show - fun-o LPL or LGPL? or both?
@1085532575 <housel!housel@cantor.gwydiondylan.org> both... see the comments for bug #6986
@1085532594 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085533600 -dircproxy- You disconnected
@1085534979 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> housel: can you send me the reg script to establish file associations or tell me where can i find it?
@1085535667 -zelazny.freenode.net- pritlove (~pritlove@port-212-202-175-14.dynamic.qsc.de) joined the channel
@1085537532 -zelazny.freenode.net- rhashem (~rhashem@ip68-100-5-48.dc.dc.cox.net) joined the channel
@1085537848 -zelazny.freenode.net- carlgay_ (~chatzilla@h00a0c5e1f619.ne.client2.attbi.com) joined the channel
@1085537862 <carlgay_!~chatzilla@h00a0c5e1f619.ne.client2.attbi.com> hi
@1085538311 -dircproxy- You connected
@1085538337 <housel!housel@cantor.gwydiondylan.org> re
@1085538365 <housel!housel@cantor.gwydiondylan.org> I haven't got a .reg file for the associations yet
@1085538401 <housel!housel@cantor.gwydiondylan.org> The FD2.0 installer puts some in the registry, you can find them in the usual place
@1085538441 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok
@1085538445 <housel!housel@cantor.gwydiondylan.org> If you can't figure it out, I'll try to work on adding them to fd.reg later this evening
@1085538455 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> btw what timezone you are in?
@1085538467 <housel!housel@cantor.gwydiondylan.org> US west coast time
@1085538499 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i think i can figure ir out - i thought you have it somewhere
@1085538524 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i am in gmt+2, so it is probably the time to sleep
@1085538558 <housel!housel@cantor.gwydiondylan.org> quite
@1085538581 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i have almost finished it, except the dual-license part - i have to concoct some kind of dialog to present the choice, and then display one or the other
@1085538608 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> alternatively, i can stuff them one after the other on the same screen
@1085538670 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> what do you think is the best?
@1085538684 <housel!housel@cantor.gwydiondylan.org> I think I'll try to write a License.txt later, that contains both of those licenses, plus the MPS license
@1085538698 <housel!housel@cantor.gwydiondylan.org> so choice 2
@1085538716 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> ok. it is also *vastly* simpler
@1085538775 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> when i finish, i'll put it up at oudeis.maclisp.org
@1085538789 <housel!housel@cantor.gwydiondylan.org> great
@1085538794 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> (if i haven't exceeded my quota there)
@1085538858 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> i tink it's ok, it will be about 90M, i have 100M left there
@1085538885 <Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net> WHOA!  It really offers free accounts on machines that run Genera??
@1085538905 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> no, not quite
@1085538919 <Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net> ...oh.
@1085538927 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> they have genera, but don't give public access to it
@1085538934 <Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net> Hrmph.
@1085538945 <oudeis!~oudeis@bzq-218-95-152.red.bezeqint.net> its openbsd on alphas
@1085538960 [Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net] ACTION is uninterested in BSD...
@1085538974 <Riastradh!~riastradh@pool-151-203-222-80.bos.east.verizon.net> (or free UNIX shells in general)
@1085539203 -dircproxy- You disconnected
@1085539238 -dircproxy- You connected
@1085539303 <housel!housel@cantor.gwydiondylan.org> I can't imagine that logging into genera over telnet would be all that enjoyable
@1085539311 <Riastradh!~riastradh@pool-151-20