Camelia, the Perl 6 bug

IRC log for #parrot, 2009-08-17

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:04 MikHel joined #parrot
00:05 dalek parrot: r40604 | NotFound++ | failed to fetch changeset:
00:05 dalek parrot: merge from trunk r40603
00:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40604/
00:08 treed Okay.
00:08 treed Smolder done.
00:09 Whiteknight joined #parrot
00:11 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r405603 - Ubuntu 9.04 amd64 (gcc)
00:11 mikehh r40603
00:17 tetragon joined #parrot
00:22 Topic for #parrotis now http://www.parrot.org | http://planet.parrot.org | 1.4.0 | Feature Freeze: No big changes before the 1.5.0 release | Current Focus: TESTING!
00:23 Whiteknight hello #parrot
00:41 dduncan joined #parrot
00:45 kid51 joined #parrot
01:13 nathanmccauley joined #parrot
01:18 Andy joined #parrot
01:26 dduncan I found a bug in Parrot's Makefile generator, at least at r40598, the Makefile generator doesn't escape spaces in the path
01:26 dduncan so build fails
01:26 kid51 Can you open a trac ticket about that?
01:26 dduncan will check newest ...
01:27 dduncan and yes I could ... may have to make an account first ...
01:35 kid51 dduncan:  In trac ticket, please include details of our OS, version of 'make', etc.
01:35 kid51 Considering the number of times that Parrot gets built each day, I am surprised at this report.
01:35 dduncan maybe no one else had spaces in their dirs
01:36 kid51 I have successfully run 'make' at HEAD (r 40604).
01:37 dduncan I have r40604 now
01:38 Whiteknight I definitely don't have spaces in my path
01:39 dduncan here's a sample line in the Makefile that Configure made for me:
01:39 dduncan ALL_PARROT_LIBS     := -L/Volumes/Documents/Documents/Muldis floss projects/git_working_dirs/rakudo/parrot/blib/lib -lparrot $(ICU_SHARED) $(C_LIBS)
01:39 kid51 dduncan:  Are you trying to build Parrot in a directory like this: C:\\my home dir  ?
01:39 dduncan no space escaping
01:40 dduncan in contrast, Perl 5 handles this properly as I recall
01:40 dduncan kid51, yes, but I'm using a Unix-compatible
01:41 dduncan sans the trailing blib/lib/, the above is where it is checked ou
01:41 dduncan sure, presumably I could rename the dir to workaround the problem, but the makefile should be robust enough to not have a problem with it
01:42 kid51 So the spaces are in your sandbox's path, not in paths built by Parrot, correct?
01:42 dduncan yes
01:43 dduncan actually I know Perl's works properly, because I also have a DBD::SQLite checkout dir under "floss projects/" and it builds/etc fine
01:43 kid51 Okay, then a TT is in order.  Although I don't have a Windows box, several of our lead developers do, so they will be in a better position than I to evaluate any patch you submit.
01:43 dduncan I don't have a Windows box either
01:43 dduncan you would have the same problem on any Unix or Linux etc
01:43 dduncan Mac OS X in my case
01:44 dduncan Unix compatible
01:44 dduncan so if you put a space in your own sandbox's dir name, you'd probably see the same issue
01:44 dduncan on the machine you have now
01:44 kid51 Okay, mention that in your TT.
01:45 dduncan its a common basic programming mistake ... shell/etc programming forgetting the files can have any chars in their names
01:46 dduncan I have to go for a bit, should add the TT within 2 hours
01:46 dduncan I will
01:54 mikehh let me test that
02:02 mikehh yup fails to build parrot in directory 'test parrot'
02:05 mikehh cc -o miniparrot src/main.o src/null_config.o \
02:05 mikehh -Wl,-rpath=/home/mhh/test parrot/blib/lib -L/home/mhh/test parrot/blib/lib -lparrot -lm -L/usr/lib  -licuuc -licudata -lm -lnsl -ldl -lm -lcrypt -lutil -lpthread -lgmp -lreadline  -fstack-protector -L/usr/local/lib -Wl,-E
02:05 mikehh cc: parrot/blib/lib: No such file or directory
02:05 mikehh cc: parrot/blib/lib: No such file or directory
02:05 mikehh make: *** [miniparrot] Error 1
02:05 dalek parrot: r40605 | allison++ | branches/pcc_arg_unify/lib/​Parrot/Pmc2c/PCCMETHOD.pm:
02:05 dalek parrot: [pcc] Mark that sig object is no longer needed after the NCI sub takes a
02:05 dalek parrot: copy for parameters and returns.
02:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40605/
02:13 kid51 mikehh:  Once dduncan creates that ticket, feel free to take it and fix it! :-)
02:19 TiMBuS joined #parrot
02:41 janus joined #parrot
02:52 davidfetter joined #parrot
03:09 dduncan I'm back ... now will login to trac and file ticket ...
03:10 dduncan a quick search didn't find any existing tickets on topic
03:12 mikehh dduncan: I tested this using a dir 'test parrot' and the build failed
03:13 dduncan I noticed
03:14 dduncan so, just created trac account ...
03:14 mikehh IIRC gnu make has major problems with spaces in dir/file names - I don't know about other makes - nmake?
03:16 dalek cardinal: c34b8a6 | treed++ |  (3 files):
03:16 dalek cardinal: Properly fix == with bools. (joeri++) Also adds some tests and adds test:bool to the test:basic (and therefore test:all) targets.
03:16 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/c34b8a6d0bc56e217fc2c5656b4748b4c8d46015
03:17 dduncan to fix, presumably you just have to escape, eg write \<space> or quote the path ... see what Perl's MakeMaker does
03:17 dduncan now I assume that I just pick the status flags I think most appropriate, but core devs will change them if they think different
03:17 dduncan for the ticket
03:19 dduncan question, regarding 'component', is a problem in the makefile more a 'build' or 'configure' problem?  I'll assume the first
03:19 mikehh try and set the appropriate flags - I will have a look later and probably add a comment
03:19 mikehh must get some sleep now
03:20 mikehh configure generates the makefiles
03:21 mikehh bbl - in a few hours
03:22 dduncan so I'll make it 'configure' then
03:23 dduncan wasn't sure if that 'component' meant something about Parrot's runtime configuration
03:23 dduncan but you seem to confirm it means the configure.pl/etc ... part of the build
03:43 dalek close: r97 | Austin++ | trunk/ (23 files):
03:43 dalek close: Grammar refactoring checkpoint. Symbol Resolution works. Need to fix types,
03:43 dalek close: resolve declarations.
03:43 dalek close: review: http://code.google.com/p/close/source/detail?r=97
03:44 dduncan okay, mikehh and kid51, the new Ticket 930 is mine
03:44 dalek TT #930 created by duncand++: build fails when sandbox path contains whitespace
04:32 tetragon joined #parrot
04:37 cotto good evening, #parrot
04:56 Andy joined #parrot
05:01 tetragon joined #parrot
05:03 nathanmccauley joined #parrot
05:17 dduncan left #parrot
05:18 tetragon joined #parrot
05:28 cotto chromatic, ping
05:30 chromatic pong
05:30 cotto Can you recommend some good remedial reading on continuations and CPS?
05:31 cotto The more I dig into this, the more I'm finding that I don't really have a mental model for what's going on.
05:32 cotto If wikipedia's entries on Continuations and CPS are good enough, that's where I am now.
05:36 allison cotto: there's really not a lot of good material out there, unfortunately
05:36 allison cotto: most of it's slanted toward functional languages
05:37 cotto I'm finding that.
05:38 cotto I have the basic idea down that a continuation represents the state of a program at a given point, but I'm having trouble getting an intuitive understand of how they're used and what can be done with them.
05:39 allison I've collected some papers: http://delicious.com/allisonrandal/continuations
05:40 allison (but, still slanted toward functional languages)
05:40 allison the basic idea is just a reset button
05:40 allison (but a partial one)
05:42 cotto allison, thanks.  I'll look through those links.
05:43 allison I think I'll have to end up writing some papers on the subject
05:44 cotto Fittingly, I just watched mst's "you're not good enough" talk a couple days ago.
05:47 * cotto finds it highly ironic that his desire to improve PHP would lead him to learn about continuations.
05:56 tetragon joined #parrot
06:03 jrtayloriv joined #parrot
06:04 gaz joined #parrot
06:09 jrtayloriv joined #parrot
06:12 uniejo_ joined #parrot
06:27 szabgab joined #parrot
06:41 HG` joined #parrot
06:51 gaz joined #parrot
06:54 tetragon joined #parrot
07:00 japhb joined #parrot
07:06 gaz joined #parrot
07:31 baest joined #parrot
07:52 gaz joined #parrot
08:03 gaz joined #parrot
08:37 payload left #parrot
08:42 jrtayloriv joined #parrot
08:45 masak joined #parrot
08:46 bacek_at_work cotto: try to google "call/cc for c-programmers"
08:49 dalek markdown: b90919f | fperrad++ | README:
08:49 dalek markdown: add a README
08:49 dalek markdown: review: http://github.com/fperrad/markdown/commit​/b90919f14963039cf4a8e2b3cf9b371698e543a1
08:52 GeJ Good morning everyone.
09:05 masak o/
09:05 eirik morning
09:18 mokurai left #parrot
09:24 cotto bacek_at_work, thanks!
09:25 cotto I wish I could sleep, but that's a good alternative.
09:25 eiro joined #parrot
09:26 eiro joined #parrot
09:40 jrtayloriv joined #parrot
09:41 bacek joined #parrot
09:43 bacek o hai
09:44 eiro joined #parrot
09:56 mikehh_ joined #parrot
10:24 bacek joined #parrot
11:02 Infinoid good morning
11:02 purl Here I am, brain the size of a planet, and all they say is 'Good Morning'
11:03 bacek and good moroning to you Infinoid
11:10 Infinoid hai bacek
11:23 jonathan Hi all. I'm getting segfaults when building Rakudo. It's finding the math_ops and loading it, but then segfaults when calling dynops_register
11:23 jonathan Erm, dynop_register
11:24 jonathan That's inside the dynop library itself, where it calls back into libparrot to register itself.
11:24 jonathan Anyone else seeing this issue? I'm on Win32+MSVC++.
11:24 jonathan done the usual make realclean of Rakudo, plus got clean Parrot checkout and so forth.
11:26 jonathan In my build tree, t\dynoplibs\math.t passes, fwiw
11:27 Infinoid hmm.  I'm having trouble updating rakudo, and trouble cloning it too
11:27 Infinoid error: Unable to verify pack 54212d785776bfea26a19ad365b04ecf0feb4c2b is available
11:28 jonathan Odd.
11:28 jonathan Curiously, parrot_install\bin\parrot parrot\t\dynoplibs\math.t works (that's from my Rakudo driectory)
11:29 jonathan Hmm. Even compiling that to a PBC and running with the installed Parrot works.
11:31 Infinoid ok.  cloning from git://github.com/rakudo/rakudo.git works, cloning from http://github.com/rakudo/rakudo.git seems to be missing some vital commits
11:32 * Infinoid tries to build it
11:32 bacek joined #parrot
11:33 Infinoid is this with installed parrot or locally generated parrot?
11:34 bacek installed.
11:34 purl installed is easy as well.
11:35 bacek I have same problem
11:35 bacek ==16025== Process terminating with default action of signal 11 (SIGSEGV)
11:35 bacek ==16025==  Access not within mapped region at address 0xC
11:35 bacek ==16025==    at 0x4074416: Parrot_set_p_p_kc (set.ops:461)
11:35 bacek ==16025==    by 0x411576F: runops_slow_core (cores.c:462)
11:35 jonathan Infinoid: I can reproduce it with both.
11:37 jonathan Oh ouch.
11:37 jonathan If I move the .loadlib for math_ops to be earlier (e.g. before we load perl6_ops)
11:37 jonathan Then it segfaults on loading perl6_ops.dll instead.
11:37 jonathan Again, in register.
11:41 szabgab joined #parrot
11:48 jonathan Hmm. It seems the segfault happens when we try to load more than one dynops lib.
11:50 jonathan Yup. I can reproduce the segfault with just:
11:50 jonathan .loadlib 'math_ops'
11:50 jonathan .loadlib 'perl6_ops'
11:50 jonathan .sub 'main' :main say "alive"
11:50 jonathan .end
11:50 jonathan erm, line break got lost int he paste.
11:50 jonathan *the
11:50 jonathan And it's not specific to Rakudo's dynops.
11:51 jonathan I can do
11:51 jonathan .loadlib 'math_ops'
11:51 jonathan .loadlib 'obscure_ops'
11:51 jonathan And get the exact same kind of segfault.
11:54 moritz trac it!
11:56 jonathan moritz: Aye, just seeing if I can't fix it first.
11:57 jonathan I'd be curious to know if anyone can reproduce this off Win32.
12:05 dalek TT #931 created by jonathan++: Segfault on Win32 when loading multiple op libraries
12:06 jonathan Ticketed, help most welcome.
12:08 quek joined #parrot
12:18 bacek Yay. src/runcore/main.c:1145
12:20 jonathan bacek: I fear it may be realloc'ing something, then looking at the old memory.
12:20 bacek It reallocs core->op_func_table on second pass and than uses it below.
12:21 jonathan bacek: Right.
12:21 jonathan I've got a local patch that seems to prevent the problem.
12:21 jonathan But it's a workaround more than a fix.
12:21 bacek (But I can be totally wrong (as usual (and I have to stop reading too much CS stuff)))
12:23 jonathan bacek: https://trac.parrot.org/pa​rrot/ticket/931#comment:2
12:25 bacek jonathan: you have to free old stuff. But +1 for approach.
12:25 jonathan bacek: Aye, I didn't consider that patch as something to apply.
12:25 AndyA joined #parrot
12:25 moritz somehow parrot's 'make install' is weird...
12:26 moritz when I do a make realclean; perl Configure.pl; make install
12:26 moritz it compiles parrot, but doesn't install it
12:26 bacek jonathan: of course. But it's better to create new stuff and replace old stuff in single point anyway.
12:26 moritz and when I run 'make install' again
12:26 moritz it actually installs ist
12:26 moritz s/ist/it/
12:31 nopaste "bacek" at 114.73.57.79 pasted "dyop patch for jonathan" (57 lines) at http://nopaste.snit.ch/17570
12:32 bacek jonathan: can you try nopasted patch? It's less hackish
12:34 jonathan bacek: That looks sane. Let me try.
12:36 nopaste "bacek" at 114.73.57.79 pasted "Lesser patch for jonathan" (22 lines) at http://nopaste.snit.ch/17571
12:37 bacek jonathan: second version doesn't change behaviour. It just fixes realloc issues. I'll commit it if confirm that it works.
12:37 bacek (Actually valgrind already confirmed it, but I don't trust soulless machine :)
12:38 jonathan bacek: Wait, you prefer the seocn one?
12:38 jonathan OK
12:38 bacek No, I prefer first. But release is tomorrow.
12:39 jonathan Actually that second patch is more along what I had pondered might work.
12:40 jonathan bacek: OK, trying it.
12:40 purl well, trying it is quicker than talking on IRC about not trying it ;>
12:40 jonathan purl: kicking you is quicker than drinking a beer. So?
12:40 purl i don't know, jonathan
12:41 bacek jonathan: it works! :)
12:42 jonathan result here forthcoming...
12:43 jonathan Yes, looking better.
12:43 jonathan bacek++
12:43 bacek ok. Committing now
12:44 jonathan Yay, make test works.
12:45 bacek jonathan: You are welcome! :)
12:46 dalek parrot: r40606 | bacek++ | trunk/src/runcore/main.c:
12:46 dalek parrot: [cage] Don't reuse reallocated func_table by old pointer in dynop_register. Closing TT#931
12:46 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40606/
12:47 szabgab joined #parrot
12:48 bacek jonathan: feel free to close TT#931 for more karma :)
12:49 jonathan bacek: Done
12:49 jonathan Thanks again!
12:49 dalek TT #931 closed by jonathan++: Segfault on Win32 when loading multiple op libraries
13:24 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r40606 - Ubuntu 9.04 amd64 (g++)
13:29 ruoso joined #parrot
13:38 mikehh rakudo (5637208) builds on parrot r40606 - make test/make spectest (up to 28011) PASS - Ubuntu 9.04 amd64 (g++)
13:48 rdice joined #parrot
13:59 quek left #parrot
14:10 braceta joined #parrot
14:13 Coke joined #parrot
14:13 Coke My IDE thought it would be funny if it did an "svn rm" when I said 'delete this file'.
14:14 Andy joined #parrot
14:18 mikehh Coke: that's what you get with IDE's :-}
14:23 mikehh partcl r587 builds on parrot r40606 - make test FAILs 7 tests but only 2 subtests - t/internals/pir_compiler.t - Failed tests:  1, 4 - 6 other tests exit status 1 but pass all subtests
14:23 mikehh thats on Ubuntu 9.04 amd64 (g++)
14:25 mikehh got to go to the store - my grandsons start school again tomorrow and need a couple of items - bbl
14:26 quek joined #parrot
14:28 jsut|work joined #parrot
14:31 Coke mikehh: I'd be interested to see the failure report for t/internals/pir_compiler.t
14:34 quek left #parrot
14:44 Psyche^ joined #parrot
14:53 PerlJam seen pmichaud
14:53 purl pmichaud was last seen on #parrot 3 days, 23 hours, 24 minutes and 53 seconds ago, saying: muixirt: I don't know what all will be possible there.  Currently the biggest stumbling block is that lexicals can only be PMCs, not int/string/num  [Aug 13 15:22:14 2009]
14:55 hercynium joined #parrot
14:57 nopaste "bacek" at 114.73.57.79 pasted "partcl failures for Coke++" (39 lines) at http://nopaste.snit.ch/17572
14:58 bacek Coke: it doesn't fail so badly on my box. But 1 and 4 are failing.
15:05 HG` joined #parrot
15:12 dalek rakudo: 45c6dd3 | masak++ | src/parser/grammar.pg:
15:12 dalek rakudo: [src/parser/grammar.pg] now parses #=[...]
15:12 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​5c6dd35250d2689f27b6aed2e9368066398358f
15:18 Coke bacek, ah.
15:33 Coke bacek_at_work, bacek, mikehh: fixed.
15:33 dalek partcl: r588 | coke++ | trunk/t/internals/pir_compiler.t:
15:33 dalek partcl: Explicitly load tcl_group everywhere.
15:34 dalek partcl: Used to be loaded by the tcl compiler...
15:34 dalek partcl: Seems like we should want to avoid having to do a loadlib here, and have it
15:34 dalek partcl: be available from the library load.
15:34 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=588
16:07 nathanmccauley joined #parrot
16:26 MikHel joined #parrot
16:26 nathanmccauley joined #parrot
16:53 nathanmccauley joined #parrot
17:01 mikehh partcl r588 builds on parrot r40606 - make test FAILs 6 tests exit status 1 but pass all subtests all fail with attempt to access code outside of current code segment
17:02 chromatic joined #parrot
17:02 mikehh presumably this is after they have passes the subtest?
17:03 MoC joined #parrot
17:03 theory joined #parrot
17:06 mikehh s/passes/passed/
17:09 mokurai joined #parrot
17:20 dalek rakudo: b60cef0 | jnthn++ | build/PARROT_REVISION:
17:20 dalek rakudo: Bump up Parrot revision we require to one that doesn't segfault the build on various platforms.
17:20 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/b​60cef03f12f93628b1cb320ad4aa75b9c43d52e
17:20 dalek rakudo: 17c4b9d | jnthn++ | src/ (2 files):
17:20 dalek rakudo: Make .HOW on a role give back the metaclass, rather than pun the role and give back the metaclass of the punned class. Correcting this also required fixing something that relied on the old broken behavior.
17:20 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​7c4b9d73f59dbde3062c8517f200cd05ac3d8a1
17:20 dalek rakudo: d77baf0 | jnthn++ | t/spectest.data:
17:20 dalek rakudo: Add S14-traits/package.t to the spectests.
17:20 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​77baf0caaf59b0632bfa1ba0098086b588d9c09
17:20 dalek rakudo: 2f177d4 | jnthn++ | :
17:20 dalek rakudo: Merge branch 'master' of git@github.com:rakudo/rakudo
17:20 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​f177d4c066b6e55e32fe5c4bfc198069c6f1af1
17:23 braceta left #parrot
17:26 dalek rakudo: 06d27c0 | jnthn++ | src/pmc/p6opaque.pmc:
17:26 dalek rakudo: Re-work the decision process for punning. Un-regresses Enum.pick.
17:26 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​6d27c0ac3c027884da36dab5ecbb04b0c20f991
18:00 joeri joined #parrot
18:07 particle (31-4.5-.9)*1.093
18:07 purl 27.9808
18:10 HG` joined #parrot
18:11 particle 22*1.093
18:11 purl 24.046
18:14 jan joined #parrot
18:18 hercynium joined #parrot
18:31 particle 23.5*1.093
18:31 purl 25.6855
18:32 * Coke buys particle a calculator.
18:32 moritz 2**128
18:32 purl 3.40282366920938e+38
18:33 Coke 2**2**2
18:33 purl 16
18:33 Coke polyglotbot: tcl: puts hi
18:33 polyglotbot Coke: OUTPUT[Parrot VM: Can't stat languages/tcl/tcl.pbc, code 2.␤main: Packfile loading failed␤]
18:33 Coke seen mdiep
18:33 purl mdiep was last seen on #parrot 210 days, 10 hours, 9 minutes and 27 seconds ago, saying: err... bed  [Jan 19 08:18:08 2009]
18:44 japhb joined #parrot
19:30 NotFound joined #parrot
19:34 particle1 joined #parrot
19:36 l3t0 polyglotbot: parrot: print "o hai\n"
19:36 * l3t0 fails
19:59 Aisling joined #parrot
20:05 Coke polyglotbot: pir: say "what?"
20:06 Coke rakudo: say "what?"
20:06 polyglotbot OUTPUT[Parrot VM: Can't stat languages/perl6/perl6.pbc, code 2.␤main: Packfile loading failed␤]
20:06 rdice joined #parrot
20:10 MikHel joined #parrot
20:59 hercynium joined #parrot
20:59 dalek rakudo: 15abd0f | jnthn++ | src/pmc/p6opaque.pmc:
20:59 dalek rakudo: Complete implementation of 'is hidden' so it actually influences deferal.
20:59 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​5abd0ff48eb9e9ca884fb4d900e89fb97cba356
20:59 dalek rakudo: d66f569 | jnthn++ | src/pmc/p6opaque.pmc:
20:59 dalek rakudo: Fully implement hides trait modifier's semantics.
20:59 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​66f5698a208956d20872d4118f846986cbd6c1d
21:05 dalek rakudo: 55c5fa1 | jnthn++ | src/pmc/p6opaque.pmc:
21:05 dalek rakudo: Corrections to the last patch, which broke a couple of spectests.
21:05 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/5​5c5fa18bb36b06cf3cef0d3c329e2c7313645d0
21:26 Whiteknight joined #parrot
21:32 bacek joined #parrot
21:38 joeri left #parrot
22:02 Limbic_Region joined #parrot
22:12 Whiteknight joined #parrot
22:36 Coke I'm not seeing a lot of segfaults. I wonder if the rakudo bug that was just fixed was the cause.
22:36 jonathan Rakudo bug?
22:36 purl hmmm... Rakudo bug is mailto:rakudobug@perl.org
22:37 jonathan Coke: The multiple dynops one?
22:37 Whiteknight don't go talkin'bout no bugs up in 'ere!
22:37 jonathan That probably mighta hit anyone using more than one dynop library at a time. :-)
22:37 Whiteknight we don't need no stinkin' bugs
22:38 dalek parrot: r40607 | allison++ | branches/pcc_arg_unify/src/call/pcc.c:
22:38 dalek parrot: [pcc] Better handling of invocant argument type in signature strings.
22:38 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40607/
22:38 allison joined #parrot
22:39 chromatic Coke, were you loading multiple dynops libraries?
22:39 rg joined #parrot
22:44 bacek good morning
22:44 purl Here I am, brain the size of a planet, and all they say is 'Good Morning'
22:45 Coke chromatic: not intentionally.
22:46 * Coke looks forward to un-skipping a ton of tests.
22:46 ascent joined #parrot
22:46 * Whiteknight doesn't even really know how the dynop loading and dispatching mechanism works
22:47 Whiteknight of course, I'm not looking at it until after the release in any case
22:47 chromatic It's ripe for refactoring.
22:47 Coke chromatic: ... is there a part of parrot that is not?
22:47 Whiteknight Coke beat me to it!
22:47 chromatic Several parts are pretty decent.
22:47 Coke usually does. OH!
22:48 Whiteknight chromatic: those are the several parts that have been refactored recently
22:48 Coke chromatic: ... is there a part of parrot that you haven't already refactored that is not?
22:48 Coke "too slow."
22:48 chromatic Can't think of one offhand.
22:49 jimk joined #parrot
22:50 Whiteknight I have very high hopes for Parrot in the future, with some major cleanup throughout it could become a pretty fantastic program
22:50 Whiteknight it's nice now, but it could become awesome
22:50 chromatic PGE and NQP/PCT are the nicest parts.
22:51 Whiteknight no question
22:51 purl rumour has it no question is too silly to ask
22:52 bacek We just need some decent optimiser for PCT-generated code...
22:52 Whiteknight MMD is nice but needs to be optimized, IO is pretty nice in general.
22:52 Coke bacek: oh sure, shun tcl again. =-)
22:53 bacek State-of-art GC! O, wait...
22:53 chromatic IO isn't bad.
22:53 chromatic We're good at code generation as part of the build.
22:54 bacek Coke: tcl? what tcl? :)
22:55 cotto Whiteknight, I'm planning on merging all branches into trunk in preparation for the release.  Any objections?
22:56 bacek cotto: What??? Merging before release???
22:56 Whiteknight cotto: merge all of them, ESPECIALLY the ones from circa 0.0.5 that don't even compile
22:57 cotto especially those
22:57 * cotto gets busy
22:57 Whiteknight chromatic: we're decent at code generation, but a lot of the generated code is needlessly redundant and non-performant
22:58 Whiteknight see, for example, the NCI thunks or the auto-generated VTABLEs for all the PMC types
22:58 chromatic No kidding.
22:58 chromatic The PCC rewiring will help tremendously.
22:59 Coke is there anything anyone can do to unblock that branch?
22:59 Whiteknight And JIT is a sin against computers, Kernighan Be Praised!
22:59 Coke JIT? what JIT?
22:59 chromatic Just refactoring PCC so that we don't have to insert ~135 lines of code in PMC .c files for each METHOD is a win.
23:00 Whiteknight what is the status of that branch now? I haven't been keeping up with it
23:00 Whiteknight I know allison is still working on it actively, but I don't know how far she is
23:00 Whiteknight (and I don't want to keep harassing her about it)
23:00 * jonathan is hoping to ask for status update on that at #ps tomorrow
23:01 jonathan I'm planning a big re-working on Rakudo's calling-related stuff in the near future.
23:02 Whiteknight another reworking of the Rakudo calling-related stuff?
23:04 bacek This is tradition :)
23:04 cotto parrot: r40609 | cotto++ | trunk (149 files):
23:04 cotto parrot: [pmc2c] merge the lazyrakudo branch back into trunk
23:04 cotto parrot: review: https://trac.parrot.org/parrot/changeset/40609/
23:04 RobertLJ joined #parrot
23:05 bacek cotto: !!!! :)
23:06 chromatic Heh.
23:06 jonathan omfg
23:06 jonathan oh, phew
23:07 * jonathan worried until he spotted that wasn't dalek...
23:07 cotto you're welcome
23:08 Coke cotto++
23:10 bacek jonathan: btw, can you drop lazyrakuo branch? It's probably useless now...
23:12 Whiteknight bacek: what's the status of the tt795_* branch, you have a completion estimate on that?
23:12 bacek Whiteknight: it's done
23:13 bacek I even tested it with partcl and rakudo (and mikehh++ helped with it)
23:13 Whiteknight oh awesome, I didn't realize it had gone so quickly!
23:13 cotto The C2 wiki is very helpful sometimes:
23:13 cotto "Call/CC is best used for blowing students' minds, or providing elegant solutions to difficult problems."
23:14 bacek Whiteknight: it was easy.
23:14 bacek cotto: I've spent 3 days groking call/cc :)
23:15 cotto Crap.  That means it'll take me at least a week.
23:16 cotto I'm getting a glimmer of its awesomeness, though.
23:16 Whiteknight I've never even heard of call/cc
23:16 Whiteknight chromatic/cotto: what's the status of the pluggable runcores branch?
23:18 dalek cardinal: 7f1a943 | treed++ | src/parser/grammar.pg:
23:18 dalek cardinal: Floats need to have at least one digit prior to the radix.
23:18 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7f1a943ec139c5d3500f4cc6b7d6e02ebf1e7598
23:18 dalek cardinal: 298dbf0 | treed++ | src/parser/grammar.pg:
23:18 dalek cardinal: foo++ and foo-- don't exist in Ruby
23:18 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/298dbf055d6fb5909c60506caa96bef5b0f39edc
23:18 nopaste "bacek" at 114.73.54.219 pasted "call/cc example for cotto++" (7 lines) at http://nopaste.snit.ch/17573
23:18 cotto Whiteknight, chromatic's part is solid and the actual profiling code is pretty good, but we need to spend some time figuring out how to map between CPS and the format that Callgrind expects (straight call/return).
23:19 Whiteknight cotto: so what is the purpose of that branch, to make runcores pluggable AND to add a profiling core?
23:19 cotto That's what it's become.
23:19 bacek cotto: meditate half an hour on nopasted code. It's actually very simple :)
23:19 Whiteknight okay
23:19 jonathan bacek: Yes, that branch can be killed.
23:20 jonathan svn rm https://svn.parrot.org/parrot/branches/lazyrakudo/ should do it, yes?
23:20 bacek jonathan: yes
23:21 jonathan Dobre.
23:21 jonathan It be goneth.
23:24 dalek parrot: r40608 | jonathan++ | branches/lazyrakudo:
23:24 dalek parrot: Remove lazyrakudo branch - we'll probably take a different approach, and this is way too old to merge now.
23:24 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40608/
23:25 bacek $dayjob time. See you
23:29 dalek cardinal: 7ae0541 | (Joeri Samson)++ | src/ (2 files):
23:29 dalek cardinal: Remove postfix ++ and --
23:29 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7ae0541fa6fd8f8a9536ed3f67d7713dca5347bd
23:34 dalek parrot: r40609 | allison++ | branches/pcc_arg_unify/src (2 files):
23:34 dalek parrot: [pcc] Fix test failures due to call signature being collected when it's
23:34 dalek parrot: still needed, and lingering on to interfere with later calls after it's
23:34 dalek parrot: not needed anymore. (The gc_register_pmc should be unnecessary after
23:34 dalek parrot: contexts properly mark their current_sig struct member. That may have to
23:34 dalek parrot: wait until contexts become GC-able.)
23:34 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40609/
23:35 dalek cardinal: 6061f7e | (Joeri Samson)++ | t/gc.t:
23:35 Coke SotD.
23:35 dalek cardinal: Use real true and false when testing GC
23:35 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/6061f7e02eef32f0ea94fa2f1e2d3c37c8210254
23:35 dalek cardinal: 0560029 | treed++ | src/classes/GC.pir:
23:35 dalek cardinal: Actually have GC's methods return true and false, rather than strings.
23:35 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/0560029ccd16ebee006b62937f9717ba8c59e101
23:45 cotto allison, wasn't the corevm make target from pcc_arg_unify?
23:46 allison cotto: yes, that's where I developed it
23:46 allison cotto: (though I never checked it into that branch)
23:46 cotto ok.  I was surprised not to see it there.
23:46 dalek cardinal: 171e24b | (Joeri Samson)++ | src/parser/grammar.pg:
23:47 dalek cardinal: Make all proto ops look the same
23:47 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/171e24bc6fc3a8840be6b1bb57834384a9a5a759
23:47 dalek cardinal: 4ca9cd2 | (Joeri Samson)++ | src/ (2 files):
23:47 dalek cardinal: Add support for () in expressions
23:47 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
23:47 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/4ca9cd219b748321d11dfabde356af3cf57bab21
23:47 dalek cardinal: 37af6e9 | fperrad++ | build/Makefile.in:
23:47 dalek cardinal: fix build with make
23:47 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
23:47 allison cotto: I did it because I was stuck in building PGE, but really wanted to be able to run the tests for a better handle on the problems
23:47 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/37af6e91727faa76600c251ffa8cd6cff1a24328
23:48 allison cotto: I try not to apply any changes on both trunk and branch because of merge nightmares, but it's probably about time for a fresh branch from trunk
23:49 Coke I hear that's easier with git. =-)
23:50 Coke (I am not advocating git. I'm merely being a jerk.)
23:50 treed It's not as easy as you might think, once you've published your branch.
23:51 treed You start getting "non-fast-forward" warnings.
23:51 Coke hurm. only some of my segfaults have mysteriously disappeared. Others mysteriously remain.
23:54 nathanmccauley joined #parrot
23:57 dalek partcl: r589 | coke++ | wiki/SpecTestStatus.wiki:
23:57 dalek partcl: update skip information.
23:57 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=589
23:57 dalek partcl: r590 | coke++ | trunk/config/PARROT_VERSION:
23:58 dalek partcl: bump parrot revision; fewer sefaults up here.
23:58 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=590
23:58 allison Coke: :) people say that, and then I see them agonizing over their latest trouble with git on IRC.
23:58 Coke Oh, I've been agonizing over git-svn for weeks now.
23:58 Coke I hear that if you leave off the '-svn' bit, it's better. I'm skeptical. =-)
23:59 * allison laughs so hard she chokes on her lemonade

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced