Camelia, the Perl 6 bug

IRC log for #parrot, 2010-06-15

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:02 szabgabx joined #parrot
00:24 dalek parrot: r47630 | tcurtis++ | branches/gsoc_past_optimization/docs​/pct/pattern/past_pattern_match.pod:
00:24 dalek parrot: Added docs for PAST::Pattern::Match.
00:24 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47630/
00:24 dalek parrot: r47631 | tcurtis++ | branches/gsoc_past_optimization/runtime​/parrot/library/PAST/Pattern/Match.nqp:
00:24 dalek parrot: Change the Bool method to call the "get_bool" vtable instead of vice versa.
00:24 purl dalek: that doesn't look right
00:24 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47631/
00:37 Coke iwbni someone at least TODO'd our smolder failures.
00:38 Coke mikehh: did I read that right, you're seeing failures in partcl-nqp?
00:41 dalek parrot: r47632 | darbelo++ | branches/gsoc_nfg/src/ops/string.ops:
00:41 dalek parrot: Use the apropiate macro to quiet a warning
00:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47632/
00:41 dalek parrot: r47633 | darbelo++ | branches/gsoc_nfg/src/string/grapheme.c:
00:41 dalek parrot: Add a missing ASSERT_ARGS()
00:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47633/
00:41 dalek parrot: r47634 | darbelo++ | branches/gsoc_nfg/src/ops/core_ops.c:
00:41 dalek parrot: Bootstrap ops.
00:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47634/
00:46 nopaste "tcurtis" at 192.168.1.3 pasted "Coke: I get these failures for "make-test" in partcl-nqp." (10 lines) at http://nopaste.snit.ch/21256
00:46 tcurtis s/make-test/make test/
00:52 Chandon joined #parrot
00:53 Chandon Are there known issues to setting a timer to unlimited repeat (with repeat count = -1)?
00:58 whiteknight nothing specific that I'm aware of
00:58 whiteknight of course, Timers are pretty well untested, unused, unloved, and ungood
00:58 whiteknight (I made up that last word, but you get the point)
00:59 whiteknight why, have a weird failure that needs a second pair of eyes?
01:00 kid51 Yes, but are timers double plus ungood?
01:00 abqar joined #parrot
01:00 whiteknight that's an understatement
01:00 purl You don't want to see ^Pudge's overstatement.
01:04 whiteknight Chandon: msg me or email me if you have a question. I'm heading off to bed nowish. Talk to you tomorrow
01:17 Andy joined #parrot
01:27 snarkyboojum joined #parrot
02:03 JimmyZ joined #parrot
02:44 khairul joined #parrot
02:44 JimmyZ joined #parrot
02:54 janus joined #parrot
02:56 gbacon joined #parrot
03:09 khairul cotto: ping
03:15 Hunger joined #parrot
03:18 cotto hi khairul
03:21 khairul hi cotto. see privmsg
03:21 plobsing joined #parrot
03:27 ash_ plobsing: i got at last some (but not all) of the libffi changes to work
03:28 ash_ you can build the corevm of parrot, and it passes some of the basic tests now (nci is apparently used to dispatch a lot of internal calls, i just realized this today, thanks to NotFound++ )  but part of PGE doesn't build and i haven't tracked down why
03:30 ash_ I also tested it on OS X and Ubuntu, they both work fine, I am going to try it on windows soon too, hopefully once i get the rest of the bugs worked out
03:33 plobsing ash_: are you refering to NCI being used to dispatch methods? I deprecated that and mean to get around to separating the two eventually.
03:34 ash_ got ya, well that works for now with my current work
03:34 ash_ i think it is not making pbc's right, i can't tell
03:35 ash_ one sec and I can show you the error I am getting
03:39 LoganLK joined #parrot
03:41 plobsing ash_: include/parrot/nci.h:21 seems exactly backwards. have I missunderstood something?
03:42 ash_ yes, thats backwards
03:43 ash_ I was trying to figure out the right place to put the include, i still have it also in the pmc itself
03:43 ash_ #include (without a space) seems to cause a pmc try to build a file that doesn't exist
03:44 plobsing what? that's news to me. PMCs are a little screwy that way though.
03:45 ash_ if i include #include <ffi.h> or #include "ffi.h" it tries to build include/parrot/ffi.h which doesn't exist
03:46 plobsing odd. I suppose I never ran into it because I never tried to create a core PMC that was a wrapper for external functionality.
03:47 ash_ http://gist.github.com/438668 is an example of the error i get
03:47 plobsing I am a little concerned about the blurring between the frame builder and the NCI PMC. Not sure if it was all that important, but I liked the distinction.
03:49 ash_ running some of the tests on parrot itself work, it just isn't building PGE right, still haven't tracked that down
03:49 plobsing now that one truly is weird. I've not run into somethinig like that except when meddling with the PIR parser.
03:49 plobsing do NCIs work properly (old-style signatures?) multi methods use those
03:50 plobsing "fun"!
03:50 ash_ ?
03:50 ash_ is there an example you can reference i can test...
03:52 plobsing I'm trying to figure out what native PMC has multimethods that would be being called
03:53 ash_ all of the imcc tests pass with the ./parrot executable
03:54 plobsing yeah, IMCC doesn't use high-level features like that thankfully enough
03:55 ash_ got ya, i am just running through tests, looking for breaks
03:55 ash_ make test won't work because it keeps trying to build PGE
03:56 plobsing if you want a quick workaround and you haven't bumped PBC_COMPAT, you can copy it from a trunk build
03:57 ash_ is there a reason the native_pbc tests all skip all their tests?
03:57 ash_ i haven't, but i also might be behind it, i am not sure
03:58 dalek parrot: r47635 | tcurtis++ | branches/gsoc_past_optimization/runtime​/parrot/library/PAST/Pattern/Node.nqp:
03:58 dalek parrot: Remove a block that does nothing.
03:58 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47635/
03:58 snarkyboojum joined #parrot
03:59 plobsing I don't know why native_pbc tests would skip. Usually they complain loudly until someone runs the magic script to silence them. Not sure if they've ever caught an actual bug.
03:59 plobsing consider yourself lucky
04:02 ash_ hmm i think there might be something wrong with mutl's? t/pmc/multidispatch.t didn't work well
04:04 plobsing yep. there's a test for the multis on native pmcs for ya! The fix is to patch pmc2c to emit the new signatures or to emit code that doesn't rely on framebuilder (like normal methods do now)
04:04 ash_ those are some bad errors, like glibc malloc memory corruption, i need to fix my memory stuff, i was just trying to get it to work first, but i should go on ahead and do that
04:05 ash_ hmm, was that recent?
04:05 ash_ i can try to merge my changes against the trunk
04:05 plobsing ash_: was what recent?
04:05 ash_ the patch to pmc2c? or are you saying thats the right fix?
04:06 plobsing it dne yet. it is merely a way forward.
04:08 plobsing there may be alternatives, but that's where I'd begin.
04:10 plobsing btw, what compiler are you using? I get a -Wall of warnings/errors when I try to compile.
04:10 ash_ on OS X i mostly use clang 2.0, on Ubuntu I am using gcc version 4.4.3 (Ubuntu 4.4.3-4ubuntu5)
04:10 ash_ is it the most recent checkout?
04:11 ash_ i just fixed a bunch of things in linux, gcc 4.4 is a lot more strict than clang is
04:11 ash_ about some warnings and errors
04:11 plobsing checked out of your github ~30 minutes ago
04:13 ash_ that sounds like the most recent
04:13 ash_ what are you using?
04:13 plobsing gcc 4.5
04:14 ash_ i'll build that on one of my computers to try it, just whatever is the latests stable?
04:15 plobsing no idea. I just use what my distro pushes out.
04:15 plobsing ^lazy^
04:15 ash_ which distro do you use?
04:16 plobsing arch linux
04:16 purl arch linux is not as cool as demi linux.
04:16 plobsing haters gonna hate
04:18 plobsing check http://nopaste.snit.ch/21257 for the build failures
04:22 plobsing apparently I can't tell the difference between 'git fetch' and 'git pull'
04:23 ash_ i think i fixed most of those
04:24 ash_ i wonder if there is an easy way i can add gcc 4.5
04:26 plobsing there's probably an ubuntu repo with it out there somewhere (and debian alternates is pretty nifty for that kind of thing)
04:26 plobsing probably not worthwhile though. when it is ready for prime time, there will be more than enough scrutiny by a wide variety of compilers.
04:31 dalek parrot: r47636 | tcurtis++ | branches/gsoc_past_optimization/docs​/pct/pattern/past_pattern_node.pod:
04:31 dalek parrot: Start adding docs for PAST::Pattern::Node.
04:31 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47636/
04:31 ash_ hmm, now i am getting a failed assertion 's'  in hash.c
04:31 ash_ i haven't gotten that before
04:32 plobsing yeah. I saw that. that looks nasty. Initialization bugs are odd beasts.
04:35 contingencyplan joined #parrot
04:35 ash_ i just merged with leto's master branch, so i'll see what that changes
04:35 plobsing somehow "CONST_STRING" has been corrupted
04:36 contingencyplan joined #parrot
04:37 ash_ what are the differences between miniparrot and parrot?
04:39 ash_ nope, still get the same error btw, with the merged trunk, but hey, everything still built the exact same as it did before
04:40 plobsing ash_: basically the only difference is that miniparrot is config-less
04:40 ash_ got ya
04:40 plobsing which means it doesn't have any include/lib/extension paths
04:42 ash_ hence -Iruntime/parrot/include on a bunch of the miniparrot invocations i assume
04:43 plobsing yes. otherwise it would choke on all the includes.
04:44 ash_ btw the documents are wrong and say Parrot_string_new_init is a function, its not though
04:44 ash_ pdd28_strings.pod
04:45 ash_ its "Parrot_str_new_init"
04:45 ash_ that gave me some errors at one point
04:45 plobsing it used to be. then we cleaned up the C function names to consistently use str in stead of string
04:46 ash_ its approaching midnight here, i think i'll head to bed though and try to look into these issues more tomorrow
04:49 ash_ if you have anymore comments though, i'll be glad to hear them, just msg me or email me, i got it to build without the hash error at one point, but i am getting a bit to tired to figure out whats wrong, debugging some of this is a headache
04:49 simcop2387 joined #parrot
05:04 dalek parrot: r47637 | plobsing++ | trunk/docs/pdds/pdd28_strings.pod:
05:04 dalek parrot: fix doc after 'string' => 'str' C API fixups
05:04 dalek parrot: ash_++ for spotting that
05:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47637/
05:47 snarkyboojum joined #parrot
06:17 uniejo joined #parrot
06:31 fperrad joined #parrot
06:41 fperrad_ joined #parrot
07:27 JimmyZ joined #parrot
07:32 dalek parrot: r47638 | tcurtis++ | branches/gsoc_past_optimization/d​ocs/pct/pattern/past_pattern.pod:
07:32 dalek parrot: Add doc for PAST::Pattern::patternize().
07:32 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47638/
07:41 whiteknight joined #parrot
10:24 bkuhn joined #parrot
10:31 bacek aloha, humans
10:36 bacek msg whiteknight It's from "Modern C++ design" book. Sources uner MIT licence available at http://loki-lib.sourceforge.net/
10:36 purl Message for whiteknight stored.
10:40 bacek msg whiteknight http://loki-lib.svn.sourceforge.net​/viewvc/loki-lib/trunk/src/SmallObj​.cpp?revision=1068&amp;view=markup It's quite similar to current PMC_Allocator without linked-list of free objects.
10:40 purl Message for whiteknight stored.
10:52 Myhrlin joined #parrot
11:05 mberends joined #parrot
11:10 lucian joined #parrot
11:56 gerd joined #parrot
11:58 gerd The announced time to release Parrot 2.5.0 has arrived! I want so switch the version number now. Is it okay?
11:58 whiteknight joined #parrot
11:59 whiteknight good morning, #parrot
11:59 somebody_ joined #parrot
12:03 gerd The Release 2.5.0 is commited in Revision 47639.
12:03 whiteknight gerd++
12:10 dalek parrot: r47639 | gerd++ | trunk (10 files):
12:10 dalek parrot: Release 2.5.0
12:10 purl Release 2.5.0 is commited in Revision 47639.
12:10 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47639/
12:15 gerd "make fulltest" is still running on my computer but I think I can tag it already?
12:15 whiteknight best to wait. The tests that cause the most problems are at the end of fulltest
12:15 whiteknight codetest, etc
12:15 pjcj joined #parrot
12:16 gerd okay I will wait until fulltest is finished
12:18 bluescreen joined #parrot
12:19 gerd fulltest just finished successful, so I think I can tag it now
12:21 gerd The commant to  tag it is executed
12:22 gerd I will start to copy the tar file to the ftp server.
12:27 dalek parrot: r47640 | gerd++ | tags/RELEASE_2_5_0:
12:27 dalek parrot: tagged release 2.5.0
12:27 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47640/
12:29 * mikehh need to reboot
12:33 whiteknight gerd++
12:39 dalek rakudo: 080700b | moritz++ | build/PARROT_REVISION:
12:39 dalek rakudo: bump PARROT_REVISION to 2.5.0 release
12:39 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​80700bfd4716fe960b1c1de8293615d04cee763
12:42 mberends Is runtime/parrot/library/postgres.pir usable? Is t/library/pg.t currently tested?
12:43 moritz mberends: it's tested, if you meet certain conditions
12:43 moritz dukeleto explained it to me once, should be in the logs somewhere
12:44 mberends ok, I'll search. A FakeDBD::Pg may be possible.
12:56 mikehh joined #parrot
12:59 lucian_ joined #parrot
13:04 mberends the postgres requirements explained from http://irclog.perlgeek.de/p​arrot/2010-05-05#i_2294653
13:05 Coke PGE is also the canary in the coalmine.
13:05 purl okay, Coke.
13:12 Coke gerd - ping me if you need someone to do the docs.parrot.org release.
13:13 Coke gerd++ #release
13:14 atrodo joined #parrot
13:24 dalek rakudo: 99b6dd0 | pmichaud++ | build/PARROT_REVISION:
13:24 dalek rakudo: Eliminate hyphen from PARROT_REVISION.
13:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/9​9b6dd0d132ac0bc850b18f540c7e05a9cb1b0b9
13:24 lucian joined #parrot
13:34 JimmyZ joined #parrot
13:35 dalek website: gerd++ | Parrot 2.5.0 "Cheops" Released!
13:35 dalek website: http://www.parrot.org/news/2010/Parrot-2.5.0
13:41 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#34409), fulltest) at r47640 - Ubuntu 10.04 amd64 (g++)
13:46 patspam joined #parrot
13:52 plobsing joined #parrot
14:01 ash_ joined #parrot
14:20 bubaflub joined #parrot
14:34 ambs joined #parrot
14:35 gbacon joined #parrot
15:03 dalek rakudo: aff0402 | jonathan++ | src/builtins/Mu.pir:
15:03 dalek rakudo: Remove a fossil.
15:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​ff0402933de94efd70833c32e4f5a78ede4591a
15:03 dalek rakudo: f78f232 | jonathan++ |  (3 files):
15:03 dalek rakudo: Switch Multi to inherit from Routine, not Code. Implement candidates on Routine
15:03 dalek rakudo: so that it just returns a list containing itself. Implement cando. Note that due
15:03 dalek rakudo: to an unrelated issue to do with augment and mapped Parrot types, .cando isn't
15:03 dalek rakudo: found on multis yet; that needs a seperate fix. Patch courtesy of (Maxim
15:03 dalek rakudo: Yemelyanov)++.
15:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​78f232838da3559dcba64ff60faef14ec850e48
15:03 dalek rakudo: 207c528 | jonathan++ | src/metamodel/Attribute.nqp:
15:03 dalek rakudo: Start to stub in attribute traits a bit.
15:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​07c5284b4a2e4a8c2e74698f9af75dfe7cd44e0
15:14 dalek rakudo: 410db0f | moritz++ | t/spectest.data:
15:14 dalek rakudo: disable some tests that now loop, due to .batch being replaced by .[^$num]
15:14 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​10db0f762809001bc9c9bf3d95efd784ac670e6
15:15 gerd joined #parrot
15:17 Topic for #parrotis now Parrot 2.5.0 Released! | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GSOC Students: trac.parrot.org/parrot/wiki/GSoCersStartHere | Priorities: fix io_ops mess in corevm/coretest, review and update documentation before release
15:17 dalek website: ash++ | GSOC NCI Updates
15:17 dalek website: http://www.parrot.org/content/gsoc-nci-updates
15:20 dalek rakudo: 2721911 | moritz++ | docs/release_guide.pod:
15:20 dalek rakudo: propose Kiev as name of next release, with link to jnthn++'s blog post
15:20 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​721911c53d2961cce8407cc33a1f9675887cb65
15:53 whiteknight ash_++ # blog post
16:05 ambs_ joined #parrot
16:11 bubaflub i'm working on getting Parrot to cross-compile onto RTEMS and currently I'm building up to miniparrot.  The problem is that the architecture I'm building for and the architecture I'm running aren't the same (cross-compiling) and so I can't bootstrap stuff after miniparrot.  Would it work to first build parrot natively, grab the bootstrapped stuff (all the .pbc's) and then build as much as I can cross-compiled?
16:18 theory joined #parrot
16:19 darbelo bubaflub: We are theoretically capable of cross-reading pbc.
16:19 bubaflub darbelo: ah, how theoretically?
16:20 bubaflub i.e. is it endian independent?
16:20 darbelo So, all pbc you generate with one miniparrot on the host should just work on any target.
16:20 darbelo Thing is, we don't really test it.
16:20 bubaflub haha, i'm about to
16:20 bubaflub sometime tonight
16:21 darbelo bubaflub: Really, all pbcs are stored in the native format for the platform that generated them, but they encode all informaiton to read them into the packfile header.
16:21 bubaflub darbelo: ok
16:22 ash_ joined #parrot
16:22 darbelo Reading from the same platform that wrote, is faster, but all of the other permutions should work.
16:22 bubaflub darbelo: excellent
16:23 darbelo Also, you can't "Round-trip" foreign pbcs, either. Which is why we don't have tests for this stuff.
16:23 bubaflub yeah, i mean, it's an incredibly hard thing to test
16:24 bubaflub out of directory building would help my project a bit here...
16:24 bubaflub i know there is a ticket for that somewhere
16:32 mikehh rakudo (2721911) builds on parrot r47640 - make test PASS, spectest_smolder -> #34413 (pugs r31268) FAIL - Ubuntu 10.04 amd64 (g++)
16:32 mikehh t/spec/S32-io/IO-Socket-INET.rakudo - Failed test:  3
16:32 mikehh 22 TODO PASSes in 7 files
16:35 mikehh wow running with no -optimize takes about 42% longer
16:42 kthakore_studymore joined #parrot
16:42 mberends mikehh: that IO-Socket-INET test is prone to failures because it depends on external factors. It may also randomly pass or fail the same test(s)
16:43 mikehh mberends: I am rebuilding with --optimize - I'll try running that test separately a few times
16:44 mberends mikehh: same here, with Ubuntu x86 and amd64
16:51 dalek parrot: r47641 | khairul++ | branches/gsoc_instrument (10 files):
16:51 dalek parrot: make codetest happy
16:51 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47641/
16:51 dalek parrot: r47642 | NotFound++ | trunk/t/pmc/stringiterator.t:
16:51 dalek parrot: test for setting wrong direction in StringIterator
16:51 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47642/
16:51 Chandon What happened to pdd02_vtables.pod ?
16:59 Coke Chandon: ?
16:59 mikehh Chandon: no idea - when was it there?
17:00 whiteknight Chandon: some extremely old pdds were deleted because they were no longer needed
17:01 Coke it was moved to pdd17.
17:02 Coke see TT#377
17:03 Coke gerd: you should join parrot-users.
17:04 * Coke sees that gerd's message is no longer in the queue.
17:05 Chandon Just saw a reference to it in a comment and was disappointed when it wasn't real. I'll update the reference to pdd17.
17:07 dalek parrot: r47643 | NotFound++ | trunk/src/pmc/stringiterator.pmc:
17:07 dalek parrot: minor cleanup of StringIterator set_integer_native, no functional changes
17:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47643/
17:08 whiteknight purl msg bacek we need to rewrite PDD 09 to match the new and expected changes. I'll try to put together a patch tonight, let me know if you have any input
17:08 purl Message for bacek stored.
17:08 ash_ what is the private2 flag used for?
17:09 whiteknight ash_: depends on the PMC. the private flags are for use by the PMC if needed
17:09 whiteknight In general, I think they are implicitly deprecated. No new PMCs make any use of them
17:09 ash_ nci uses them, i am not sure what they do
17:09 whiteknight ash_: you're going to have to read the NCI PMC then to find out what it does.
17:11 whiteknight every PMC type can use them differently
17:11 ash_ got ya
17:17 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#34417), fulltest) at r47641 - Ubuntu 10.04 amd64 (g++ with --optimize)
17:18 arnsholt joined #parrot
17:23 dalek parrot: r47644 | Chandon++ | trunk/src/pmc/timer.pmc:
17:24 dalek parrot: Remove comment reference to archaic pdd.
17:24 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47644/
17:37 Coke what, who let gsoc students commit things in trunk! ;)
17:38 Coke Chandon++
17:40 dalek parrot: r47645 | Chandon++ | branches/gsoc_threads (55 files):
17:40 dalek parrot: [gsoc threads] sync with trunk
17:40 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47645/
17:52 dalek website: darbelo++ | Progress, refactorings and tests.
17:52 dalek website: http://www.parrot.org/content/p​rogress-refactorings-and-tests
17:56 mikehh rakudo (2721911) builds on parrot r47641 - make test PASS, spectest_smolder -> #34418 (pugs r31268) PASS - Ubuntu 10.04 amd64 (g++ with --optimize)
17:56 mikehh 22 TODO PASSes in 7 files
18:01 clinton joined #parrot
18:03 mikehh not really a benchmark but at the same revions parrot/rakudo 1) g++, 2) g++ with --optimize
18:03 mikehh Files=475, Tests=35629, 1669 wallclock secs (19.29 usr  3.25 sys + 5869.78 cusr 89.26 csys = 5981.58 CPU)
18:03 mikehh Files=475, Tests=35629, 883 wallclock secs (20.23 usr  3.32 sys + 3017.62 cusr 82.85 csys = 3124.02 CPU)
18:03 mikehh both spectest_smolder with TEST_JOBS=5
18:11 particle which g++ ?
18:12 lucian joined #parrot
18:13 mikehh particle:  Ubuntu 9.10 i386/amd64 - g++ (Ubuntu 4.4.3-4ubuntu5) 4.4.3
18:14 mikehh sorry those tests were today on amd64 - g++ (Ubuntu 4.4.3-4ubuntu5) 4.4.3
18:17 mikehh I get similar results on i386
18:23 mikehh and that is Ubuntu 10.04 NOT 9.10
18:28 particle all: if there's any opposition to the possibility of releasing a 2.5.1 specifically to add features in support the first r* release, please make it known
18:32 mikehh particle: maybe you should bring it up in #ps (2 hours) - I don't see a problem
18:33 particle mikehh: i will
18:43 Coke particle: why would anyone oppose that.
18:43 Coke that's just asking for trouble.
18:43 particle coke: beats me
18:43 Coke If there's a showstopper that would impact R*, we'll cut another release. NBD.
18:45 particle it's likely to be nqp-rx changes
18:45 tewk joined #parrot
18:46 whiteknight particle: is there a list of such necessary changes?
18:46 Coke I can't imagine there's a list or we would have heard about it already.
18:49 mattp /win 1
18:50 mikehh Coke: partcl PASSes make test, partcl-nqp FAIL
18:50 kthakore_studymore mattp /lose 2
18:51 particle looks like backtracking subrules and better charclass support in nqp-rx
18:51 particle whiteknight: about 2 days of pmichaud's time
18:52 whiteknight ok. 2 days of his time is certainly nothing to ignore. If they need it, I say cut the release
18:54 pmichaud_ just to be clear, the R* delays relating to Parrot have nothing to do with shortcomings of Parrot itself.  Or, to the extent they're shortcomings, they're my own.  :)
18:54 pmichaud_ I'm extremely happy with Parrot at the moment.  :)
18:54 pmichaud_ having been away from active Rakudo development for a while, and then coming back in May, I'm, well, amazed at the performance improvements that happened over that period of time.
18:56 whiteknight pmichaud: Ive been saying for months, Parrot will get better (even if it occasionally gets worse first)
18:59 bubaflub sometimes gotta work through those local minimums
19:01 cognominal "shortcomings, they're my own. ".  Problem is :  pmichaud_++ is only human.
19:02 pmichaud_ cognominal: I agree, that's a problem.  :-)  But I'm not looking for a solution to that one these days.  :)
19:03 particle pls clone pmichaud
19:03 particle sudo pls clone pmichaud
19:03 particle damned machine.
19:04 PerlJam particle: we've been trying to cloen him for years; unfortunately he's very resistant to cloning.
19:04 particle i think he doesn't want to be called "dolly"
19:05 cotto_work What if he *is* the clone?
19:05 cotto_work </twist ending>
19:07 cognominal Am I understand it that well? Is Lorito to Parrot what microcoding is to physical processors?
19:08 cotto_work That's pretty close to the truth.
19:15 Coke pmichaud_: you are eligible for a complimentary mandatory upgrade. </cyberman>
19:15 chromatic joined #parrot
19:20 khairul joined #parrot
19:28 ash_ okay
19:28 ash_ i am confused by this
19:32 ash_ https://svn.parrot.org/parrot/​trunk/src/interp/inter_misc.c line 81-94, it was doing something wrong there, it was making an NCI object with no signature. So In my branch I added a signature to it "vJP", which seemed to work yesterday and now today it seems to die during the build process, but only sometimes http://github.com/ashgti/parrot/blob/​gsoc_nci/src/interp/inter_misc.c#L86 is my change. It seems to give a hash key error when it builds
19:32 ash_ on occasion, but adding in anything on line 88 (like int i = 1 + 1;) makes it stops failing and so far doesn't stop working
19:33 chromatic Sounds like a memory problem.
19:35 snarkyboojum joined #parrot
19:35 ash_ i am going to build it with clang to see if it has a different result
19:36 dalek parrot: r47646 | NotFound++ | trunk (2 files):
19:36 dalek parrot: some code cleanup and a new test in Eval PMC to improve coverage, no functional changes
19:36 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47646/
19:36 sorear ash_: sudo sh -c 'echo 0 > /proc/sys/kernel/randomize_va_space' and try again a few times
19:37 ash_ building with clang it dies in the same chunk of code it used to, it just fails an assertion that gcc 4.2 didn't seem to execute
19:37 ash_ sorear: i am no OS X, i don't think that will work :P
19:37 NotFound ash_: What assertion?
19:37 purl assertion is there to make sure that pos stops one space short of the end of the whitespace.
19:37 ash_ key_hash_STRING assertion on s
19:38 ash_ line 151 of src/hash.c
19:39 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#34421), fulltest) at r47645 - Ubuntu 10.04 amd64 (gcc with --optimize)
19:39 ash_ i wonder why gcc doesn't fail the assertion
19:39 NotFound ash_: Have you tried a make realclean and rebuild? CONST_STRING sometimes gets confused, maybe because of a missing dependence.
19:40 ash_ i have, but i'll re-checkout my repo to be sure
19:42 Tene sorear: fwiw, for the case of sudo redirects, I do like: echo 0 | sudo tee /proc/sys/kernel/randomize_va_space
19:42 Tene then you can tab-complete the path
19:42 sorear Nice trick.
19:45 LoganLK joined #parrot
19:47 ambs_ joined #parrot
19:47 ash_ yup, i get the same segfault
19:48 ash_ adding "int i = 1 + 1;" makes it work though
19:48 * ash_ baffled
19:48 whiteknight ....
19:48 chromatic Time for valgrind.
19:49 ambs valgrind++
19:49 chromatic Unless it's obviously an uninitialized variable.
19:49 atrodo ash_> looks sort of like my issue
19:50 ash_ so, its probably something to do with CONST_STRING?
19:50 atrodo set_pmc_keyed_str() on a hash will randomly work, randomly tell me "set_pmc_keyed_str() not implemented in class 'Continuation'", where Continuation can be any PMC i've used up to that point
19:51 chromatic That sounds like invalid PBC.
19:52 atrodo It's embedded code, not a PBC
19:53 sorear chromatic++ being more resistant to verbal attack than _why
19:54 chromatic Being able to prove his assertions crazy helps.
19:56 ash_ i changed mine from the CONST_STRING macro to string_make with the const flag set
19:56 ash_ making, i'll see if that helped
19:57 ash_ nope, seg fault again
19:58 NotFound ash_: I tried your change to register_raw_nci_method_in_ns, but doesn't build with the "vJP" signature. It does if i change it to "vJ"
20:00 ash_ hmm maybe i have the signature wrong, i thought all of the METHOD pmc functions ended up with the same C signature, void Parrot_PMCName_nci_methodname(PARROT_INTERP, PMC * _self);
20:01 ash_ oh, nope, i was wrong
20:01 ash_ mutli's look different
20:01 ash_ dang
20:01 whiteknight damnit
20:01 ash_ well, that changes things
20:02 NotFound ash_: there is a special case for zero arguments (and J doesn't count as one), in that case number of arguments isn't checked.
20:02 NotFound But is a bug, not a feature, better don't depend on that.
20:03 whiteknight arguments for methods are all handled by PCC. At least, they should be
20:03 smash joined #parrot
20:04 smash hello everyone
20:04 whiteknight multis may be different, I think plobsing mentioned something about that
20:04 whiteknight hello smash
20:04 Psyche^ joined #parrot
20:04 ash_ well, plobsing said the right way was to make these raw functions not use NCI at all
20:05 NotFound With that signature change all test pass for me.
20:05 ash_ to just vJ ?
20:06 tcurtis joined #parrot
20:06 NotFound ash_: yeah.
20:06 ash_ still, is that right? My nci system is different than trunk, I have it making all of the thunks dnyamically with libffi
20:07 NotFound ash_: I just applied your change to register_raw_nci_method_in_ns to trunk.
20:07 ash_ so having the right signature is more important in my branch because thats how libffi knows what to make
20:08 NotFound ash_: but if pcc throws because of the wrong number of arguments, it can't work with the vJP signature.
20:09 NotFound ash_: so I doubt that that simple change can work without lots of workarounds.
20:12 mikehh rakudo (2721911) builds on parrot r47645 - make test PASS, spectest_smolder -> #34423 (pugs r31268) PASS - Ubuntu 10.04 amd64 (gcc with --optimize)
20:12 mikehh 22 TODO PASSes in 7 files
20:12 ash_ hmm changing the signature to "vJ" gives http://gist.github.com/439660 on the part where it always fails
20:13 Andy joined #parrot
20:14 mikehh #ps in 15
20:15 darbelo tcurtis: nopastes expire. Can you put that up somewhere more lasting_
20:24 atrodo going to YAPC::NA 2010?
20:24 purl i guess going to YAPC::NA 2010 is qw(jhannah rbuels cfedde apeiron dha nacmac dhoss mst chargrill kyriel triddle DrForr coke packy kolibrie colomon jasonmay arcanez dylan doy ology waltman autarch pmichaud atrodo frew samy)
20:25 Coke pmichaud_: when are you due to arrive?
20:26 pmichaud_ 3pm Sunday, I think
20:26 * pmichaud_ checks
20:26 * Coke sees http://yapc2010.com/yn2010/wi​ki?node=ArrivalsAndDepartures, that looks newish.
20:26 pmichaud_ arrive 3:10p on Sunday
20:27 particle no, going to YAPC::NA 2010 is qw(jhannah rbuels cfedde apeiron dha nacmac dhoss mst chargrill kyriel triddle DrForr coke packy kolibrie colomon jasonmay arcanez dylan doy ology waltman autarch particle pmichaud atrodo frew samy)
20:27 purl okay, particle.
20:27 particle there, it's slightly more sorted now
20:28 atrodo I like the note on the ArrivalsAndDepartures page> "no train service since April 28, 1977"
20:28 atrodo I can't believe someone actually looked that date up
20:31 bluescreen joined #parrot
20:31 chromatic #ps time
20:34 particle Coke: where you staying for yapc?
20:34 Coke dorm.
20:34 Coke con packy.
20:34 eternaleye joined #parrot
20:34 particle i booked too late for either dorm or con hotel, so i'm in the hyatt or something
20:36 tcurtis darbelo: http://gist.github.com/439694
20:38 allison joined #parrot
20:39 hercynium joined #parrot
20:39 * allison was helping out at a school party for son
20:39 Tene looks like no yapc::na for me this year.  maybe next year.
20:48 Topic for #parrotis now Parrot 2.5.0 Released! | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GSOC Students: trac.parrot.org/parrot/wiki/GSoCersStartHere | Priorities: merge gc_massacre, remove deprecated items, close tickets.
20:54 kthakore_studymore win 4
20:56 cotto_work win win win win
20:56 kthakore_studymore cotto_work: you win 4
20:56 kthakore_studymore hi cotto_work
20:56 NotFound xed
20:56 kthakore_studymore hi NotFound
21:02 preflex joined #parrot
21:08 theory joined #parrot
21:13 whiteknight joined #parrot
21:17 jan joined #parrot
21:31 dalek website: Chandon++ | A New Design for Timers
21:31 dalek website: http://www.parrot.org/content/new-design-timers
21:31 dalek TT #1483 closed by NotFound++: optimization for mixed_cs_index
21:31 dalek TT #1483: http://trac.parrot.org/parrot/ticket/1483
21:31 dalek TT #1659 closed by NotFound++: FileHandle objects don't honour timely destruction
21:31 dalek TT #1659: http://trac.parrot.org/parrot/ticket/1659
21:38 whiteknight Chandon++ # blog post
21:38 whiteknight Getting timers working right is good, improving performance is better
21:43 Chandon Someone's going to hit me for threatening to break the one thing that timers can do ATM on platforms without threads or signals.
21:47 dalek parrot: r47647 | NotFound++ | trunk/t/oo/vtableoverride.t:
21:47 dalek parrot: add a reference to the ticket in the regresion test, TT #1593
21:47 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47647/
21:48 dalek TT #1593 closed by NotFound++: parrot puts an invalid value in a register when vtable override does not ...
21:48 dalek TT #1593: http://trac.parrot.org/parrot/ticket/1593
21:57 bubaflub joined #parrot
22:06 whiteknight Chandon: what?
22:07 whiteknight Chandon: a flag and #ifdef can fall back on old behavior
22:08 Chandon Yup. Which means that this thing will need to get written twice. Yay!
22:14 cotto_work Wow.  LWN has already posted the Parrot 2.5 release annoucement.
22:14 darbelo .oO(What platforms without threads or signals are we targetting?)
22:16 whiteknight Chandon: do we support any platforms without threads  or signals?
22:16 whiteknight I suggest we declare the answer to be "no"
22:16 Chandon Are any platforms supported that aren't POSIX or Win32?
22:17 bubaflub well, some RTEMS platforms might not have POSIX or threading enabled...
22:17 bubaflub but i don't think you should worry about that at this point, honestly
22:18 Chandon Now for an easy question: How do I add a .c file to the build process?
22:18 darbelo .oO(What platforms that aren't POSIX or Win32 are we targetting?)
22:18 darbelo ;)
22:18 whiteknight Chandon: makefile is in config/gen/makefiles/root.in
22:19 whiteknight Chandon: I'm sure you can find a file to copy from there
22:19 darbelo Is that a .c file that should end up linked into libparrot?
22:19 Chandon darbelo: VMS, TI calculators, Arduinos, etc.
22:20 Chandon Yea. config/gen/makefiles/root.in looks like what I want.
22:20 dolmen joined #parrot
22:26 whiteknight Configure.pl turns that into Makefile
22:27 cotto_work I'll buy anyone a pizza who can make Parrot run on a TI calculator.
22:28 cotto_work For VMS, that person can buy me a pizza.
22:30 NotFound cotto_work: putting a PC-on-a-chip inside the cage of the calculator qualifies?
22:31 darbelo I'm not going to say it's impossible, but NotFound's idea is likely to be easier...
22:32 cotto_work note that for reasons of practicality, I may end up eating said pizza on that person's behalf
22:32 NotFound Even easier, get a N900 and install Parrot and a TI emulator on it X-)
22:33 NotFound For extra points, implement the TI emulator with parrot.
22:33 cotto_work doesn't count, though I may end up buying myself a pizza anyway.
22:34 cotto_work NotFound: for that, I'd buy 1.5 pizzas
22:34 cotto_work maybe even 2
22:35 kid51 joined #parrot
22:36 cotto_work in the past, it used to be hard to find a live streaming video of people playing with cats to Devo's latest album
22:39 chromatic ... riding Roombas
22:53 dalek parrot: r47648 | NotFound++ | trunk (2 files):
22:53 dalek parrot: ByteBuffer push_integer vtable and its tests
22:53 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47648/
22:53 dalek parrot: r47649 | jkeenan++ | trunk/t/steps/gen/opengl-01.t:
22:53 dalek parrot: Correct functioning of tests on boxes where OpenGL is not really installed:  �http://trac.parrot.org/parrot/ticket/1680.
22:53 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47649/
22:54 dalek TT #1258 closed by jkeenan++: Method cache not always invalidated
22:54 dalek TT #1258: http://trac.parrot.org/parrot/ticket/1258
22:54 dalek TT #1680 closed by jkeenan++: t/steps/gen/opengl-01.t: Tests for verbose output fail when OpenGL not ...
22:54 dalek TT #1680: http://trac.parrot.org/parrot/ticket/1680
22:56 cotto_work NotFound, is there any reason for using STATICSELF instead of SELF in r47648?
22:59 NotFound cotto_work: speed
23:01 cotto_work Is that worth the potential cost of surprising behavior if someone subclasses ByteBuffer?
23:02 cotto_work (my vote is "no" btw)
23:03 NotFound cotto_work: most probably not.
23:04 eternaleye joined #parrot
23:04 NotFound Changing...
23:10 dalek parrot: r47650 | NotFound++ | trunk/src/pmc/bytebuffer.pmc:
23:10 dalek parrot: use SELF instead of STATICSELF to prevent breaking inheritance, cotto++
23:10 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47650/
23:16 bubaflub joined #parrot
23:27 Chandon I don't think headerizer likes me: Couldn't handle "PARROT EXPORT void Parrot_timers_init()"
23:28 Chandon Err... PARROT_EXPORT
23:29 cotto_work you may need those on separate lines
23:29 cotto_work istr that some part of the build requires that
23:30 Chandon It's on three separate lines now.
23:30 cotto_work is that in your branch?
23:31 Chandon It's not committed yet, but yes.
23:31 cotto_work which branch?
23:31 purl which branch is the one I need to pull again?
23:32 cotto_work I was hoping that purl would have something more clever.
23:32 Chandon gsoc_threads
23:34 cotto_work looks like headerizer is broken without your uncommitted changes
23:35 Chandon That's not good.
23:37 cotto_work it might need a built parrot
23:37 Chandon Yea, it coughs up other errors without that.
23:39 cotto_work it gripes but runs fine with a built parrot
23:39 cotto_work do you have a patch?
23:39 mikehh_ joined #parrot
23:40 Chandon Let me try to make a simple test case.
23:40 cotto_work you can nopaste a diff
23:41 cotto_work reducing a headerizer problem may not be worth the effort
23:44 nopaste "Chandon" at 192.168.1.3 pasted "svn diff in gsoc_threads ; breaks headerizer" (155 lines) at http://nopaste.snit.ch/21275
23:45 nopaste "Chandon" at 192.168.1.3 pasted "svn diff in trunk ; reduced headerizer breakage" (19 lines) at http://nopaste.snit.ch/21276
23:55 cotto_work Chandon: you need function_name(void) instead of function_name()
23:55 cotto_work That'll get you to the next error message. ;)
23:56 Chandon I like getting to the next error message.
23:57 Chandon Why is function(void) preferred to function()?
23:57 cotto_work I don't know.
23:58 cotto_work It might be a c89 thing.

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

Parrot | source cross referenced