Camelia, the Perl 6 bug

IRC log for #parrot, 2010-01-19

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 dalek parrot: r43474 | jkeenan++ | branches/no_running_make_test:
00:00 dalek parrot: Creating no_running_make_test in �https://svn.parrot.org/parrot//branches
00:00 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43474/
00:00 dalek parrot: r43475 | jkeenan++ | tags/no_running_make_test-43473:
00:00 dalek parrot: Tagging trunk at r43473 so that the no_running_make_test can later be synched to it.
00:00 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43475/
00:03 plobsing darbelo: if the failure doesn't happen in every run, is it possible you got lucky when trying to run it outside of the harness?
00:04 darbelo plobsing: yep, that was the issue.  My 'fix' didn't fix it.
00:05 darbelo And running "parrot -R gcdebug t/pmc/io_22.pir" takes too long and doesn't segfault.
00:05 darbelo At least not yet.
00:06 Andy joined #parrot
00:12 ruoso joined #parrot
00:13 mikehh joined #parrot
00:14 dalek TT #1412 created by jkeenan++: Remove 'running-make-test'
00:16 dalek parrot: r43476 | jkeenan++ | branches/no_running_make_test (5 files):
00:16 dalek parrot: Remove '--running-make-test' in all its spellings per �http://trac.parrot.org/parrot/ticket/1412.
00:16 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43476/
00:16 Khisanth joined #parrot
00:17 kid51 no_running_make_test branch passed make test and make codetest
00:18 kid51 So, unless someone speaks up for that option, it is safely deletable.
00:19 darbelo Kill!
00:19 darbelo Kill!
00:19 darbelo Kill!
00:19 plobsing darbelo: sorry about the slowness. you could try throwing in 'sweep' and/or 'collect' ops right before the failure. if it's a gc issue, those should blow it away every time.
00:22 darbelo plobsing: The problem isn't in making it blow up. It's in finding out why ;)
00:22 plobsing and step 1 is getting it to blow up every time
00:23 mtk joined #parrot
00:24 tetragon joined #parrot
00:25 darbelo Inserting "Parrot_gc_mark_and_sweep(interp, 0)" at line 661 of src/io/buffer.c should do it.
00:25 plobsing the weird thing about the gc theory, is the only place I can see the buffer being set is src/io/buffer.c:109 which does a mem_sys_allocate
00:26 plobsing darbelo: does it do it?
00:27 mtk left #parrot
00:30 darbelo nope.
00:31 darbelo I need a new theory.
00:33 plobsing try throwing in a PARROT_ASSERT(buffer_flags & PIO_BF_MALLOC)
00:34 plobsing also, what are the pointer values when they are broken? random? null? repeatable constant?
00:34 darbelo plobsing: wait. Whare are you seeing buffer malloc()ed ?
00:34 plobsing src/io/buffer.c:109
00:35 darbelo That's the wrong buffer.
00:35 darbelo see line 568
00:36 plobsing oh. I was under the impression buffer_start was the broken pointer
00:37 darbelo That's what I thought at first, but nope. it's buffer that points into never-never land. That's why I therized a pool compaction.
00:37 plobsing that one's probably easy: replace all instances of buffer with s->strstart
00:38 plobsing and maintain the offset yourself
00:40 plobsing alternatively you could block GC through that stretch, but there's a slippery slope to the entire VM having GC dissabled
00:47 darbelo Or, I could Parrot_str_to_cstring() the sucker and avoid both tickling the gc *and* poking into the string.
00:47 plobsing and have unnecessary mallocs
00:48 abqar joined #parrot
00:49 plobsing quick test: try replacing 'buffer = s->strstart' with '#define buffer s->strstart' (and a matching #undef at the end of the function)
00:49 darbelo Encapsulation costs...
00:49 Coke (new theory) it's a demon. a dancing demon --- no, something isn't right there.
00:50 nopaste "darbelo" at 190.192.220.13 pasted "The Parrot_str_to_cstring() way. No segfaults." (37 lines) at http://nopaste.snit.ch/19297
00:52 plobsing or what about having buffer point to s->strstart and dereference every access?
00:53 nopaste "darbelo" at 190.192.220.13 pasted "The s->strstart way. No segfaults either." (48 lines) at http://nopaste.snit.ch/19298
00:55 darbelo Both work, somebody should benchmark them and figure out if the speed gain is woth the encapsulation breakage.
00:55 plobsing seen somebody?
00:55 purl somebody was last seen on purl 5 years, 159 days, 12 hours, 43 minutes and 47 seconds ago, saying: <private message>  [Aug 13 12:09:03 2004]
00:55 darbelo But I can't do it now, I have to go.
00:56 hercynium joined #parrot
00:57 darbelo If nobody beats me to it, I'll callgrind it tomorow.
00:57 darbelo See y'all later.
01:19 ruoso joined #parrot
01:20 cconstantine joined #parrot
01:23 kid51_at_dinner nick kid51
01:39 cconstantine joined #parrot
01:47 cognominal joined #parrot
02:42 cconstantine joined #parrot
02:44 eternaleye joined #parrot
03:34 tetragon joined #parrot
03:59 Khisanth joined #parrot
04:21 kurahaupo joined #parrot
04:23 kurahaupo I'm at LCA2010; anyone else here in Wellington?
04:29 cotto be quiet you, with your warm weather and geeky conference
04:30 cotto (short answer:  no)
04:32 GeJ Yeah, like he said!
04:33 GeJ (well, at least for the geeky conference)
04:33 kurahaupo Quote 2 minutes ago in the round-table session on parallel programming, by the Intel guy: "we can now exceed Moore's law" followed by "we could produce a CPU with 50 core ... but nobody would buy it, so we won't"
04:34 cotto That's not quite Moore's law.
04:34 kurahaupo So getting parallism right is crucial to making Perl+Parrot viable in the long term.
04:35 kurahaupo Cotto: no, of course it's not literally Moore's law. But if you can fully automate your parallism, it will amount to the same thing in practice.
04:35 cotto So far that's proven to be a big "if", but yes.
04:36 kurahaupo There hasn't been *too* much bloody spilt between pro and anti factions regarding pure functional languages like Haskel. :-)
04:37 kurahaupo BTW I think I've figured out the basic problem with the array unshift problem.
04:38 kurahaupo The quick indexing code only works (and is supposed to be guarded against being used) when all the chunks are the same size. But something sets the policy to "fixed" even after it has variable sized chunks.
04:40 GeJ hum... my former boss had a talk at LCA2010.
04:41 kurahaupo1 joined #parrot
04:42 kurahaupo1 Gej: what's your former boss's name?
04:43 GeJ Lenz Gschwendtner.
04:45 GeJ I moved from a PHP shop to a Perl one. So I guess I have to blame him somewhat for hanging in here.
04:46 kurahaupo1 I'll ask round
04:46 GeJ try black beard, ponytail, and a strong german accent.
04:47 treed doesn't seem terribly distinctive
04:47 GeJ I bet the "strong german accent" is not so common in NZ.
04:47 treed There's a perl guy somewhore with a pony-tail?
04:47 treed Ah, maybe.
04:47 treed There's a guy at work who matches that description except for the ponytail, since he's bald.
04:48 treed And he doesn't do perl.
04:48 GeJ Lenz is doing Erlang these days... with still some Perl in it.
04:49 GeJ the slides of his talk : http://norbu09.org/2010/01/1​9/slides-from-LCA-talk.html
04:50 GeJ anyway, if you see him, tell him his french guy says "hi".
04:50 GeJ :)
04:51 kurahaupo joined #parrot
04:51 kurahaupo Lenz says hi
04:51 GeJ :)
04:51 GeJ thanks
04:52 kurahaupo Am standing outside centre close to glass so I can pick up wifi. Going for dinner...
05:36 patspam joined #parrot
05:39 cconstantine joined #parrot
05:43 cognominal joined #parrot
05:59 Tene Ouch.  I still have the weird autotools issue.
06:10 dalek plparrot: 443faf6 | (Joshua Tolley)++ | src/handler/plparrot.c:
06:10 dalek plparrot: Add code for better return value handing, and to fix a crash in the tests
06:10 dalek plparrot: review: http://github.com/leto/plparrot/commit/4​43faf656519dc3ae222e1ff01030a1418e1d4d2
06:30 dukeleto eggyknap++
06:49 chromatic joined #parrot
06:59 uniejo joined #parrot
07:00 dalek parrot: r43477 | tene++ | trunk/config/auto/icu.pm:
07:00 dalek parrot: Minor workaround for an extra newline with icu-config on my platform that breaks the build for me.
07:00 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43477/
07:02 Tene leto_: lemme know if you need more contact information for me for your mysterious somebody.
07:13 dukeleto Tene: Perry Wagle is his name wagle@mac.com
07:13 Tene Have him email me
07:13 * Tene at allalone.org
07:14 dukeleto Tene: he is wagle on freenode right now
07:14 Tene ah
07:14 dukeleto i will get him in here
07:14 Tene I'll be awake for another little bit
07:21 dukeleto i gave him your info
07:24 Tene Guess I'll chat with him tomorrow.  Sleeping now-ish.
07:24 Tene 'night all
07:33 * wagle has always been here
07:36 chromatic wagle is Kosh!
07:46 dukeleto ?
07:46 dukeleto chromatic: i have been talking to wagle about scheme on parrot
07:47 chromatic Excellent.
07:47 dukeleto chromatic: he is very interested and I told him to target 2.0 and use the newer skeleton script. have any other advice for him?
07:48 dalek parrot: r43478 | gerd++ | trunk/PLATFORMS:
07:48 dalek parrot: add ppc, ppc64 and arm architectures running with linux as new platforms
07:48 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43478/
07:49 chromatic That's the first place to start.  NQP-ng is the way to go.
07:50 chromatic NQP-rx?  Whichever it is now.
07:51 dukeleto chromatic: we are supposed to just say "the NQP that comes with Parrot core", right ?
07:52 dukeleto ;)
07:52 * dukeleto is confused too, since I have take a little hiatus
07:52 dukeleto taken, even
07:52 chromatic yes
07:52 chromatic ... but I'm not sure which one is the new one.
07:57 * wagle is kinda groggy right now
07:58 wagle dukeleto: you need to watch babylon 5
07:58 dukeleto wagle: ok ::looks innocent::
07:59 wagle not quite [erl?
07:59 wagle perl
07:59 dukeleto nqp?
07:59 purl nqp is probably http://github.com/perl6/nqp-rx
07:59 dukeleto nqp-rx?
07:59 purl it has been said that nqp-rx is http://github.com/perl6/nqp-rx
07:59 dukeleto nqp-ng?
08:00 chromatic Sounds right.
08:13 iblechbot joined #parrot
08:13 wagle so rx, not ng
08:14 dukeleto NQP-rx is the newest, NQP is what is in Parrot core, rakudo-ng is the new branch of rakudo that is not production ready yet
08:15 * dukeleto is working off of mildy old but hopefully useful and approximately correct infoz
08:18 mikehh joined #parrot
08:18 chromatic pmichaud adds NQP-rx snapshots.
08:20 dukeleto chromatic: i thought they fell from the sky due to occasionally sacrificing yacc+lex virgins?
08:22 chromatic We're enlightened.
08:48 * dukeleto likes an occasional sacrifice. usually it is a bash script
09:30 AndyA joined #parrot
09:32 kuahara left #parrot
09:39 preflex joined #parrot
09:47 dukeleto trunk smolder_test PASS on darwin x86
10:12 JimmyZ joined #parrot
10:21 cognominal joined #parrot
10:25 abqar joined #parrot
10:34 dalek plparrot: 55278d8 | dukeleto++ | t/test.sql:
10:34 dalek plparrot: Fix plan in the tests
10:34 dalek plparrot: review: http://github.com/leto/plparrot/commit/5​5278d8c56d62d1978164c51c3c6c5f815860803
11:12 iblechbot_ joined #parrot
11:28 bacek joined #parrot
11:51 cconstantine joined #parrot
12:00 krunen joined #parrot
12:14 payload joined #parrot
12:38 dalek parrot: r43479 | bacek++ | branches/gc_encapsulate/src/gc/gc_ms.c:
12:39 dalek parrot: Stub implementation of GC MS get_info
12:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43479/
12:39 dalek parrot: r43480 | bacek++ | branches/gc_encapsulate/src/gc (2 files):
12:39 dalek parrot: Implement reallocate buffer. Fix allocate buffer
12:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43480/
12:39 dalek parrot: r43481 | bacek++ | branches/gc_encapsulate/src/gc/api.c:
12:39 dalek parrot: Fix reallocating string storage.
12:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43481/
12:41 ruoso joined #parrot
12:55 fperrad joined #parrot
12:57 bluescreen joined #parrot
13:05 he joined #parrot
13:40 payload joined #parrot
13:51 whiteknight joined #parrot
14:16 particle joined #parrot
14:16 smash joined #parrot
14:16 smash hello everyone
14:37 whiteknight hello smash
14:46 Coke ~~
14:46 Coke msg chromatic I am very distracted by $DAYJOB, bug can be roused if you need help with the release.
14:46 purl Message for chromatic stored.
14:47 * smash All tests successful. (HEAD revision)
14:48 mikehh no_running_make_test branch ALL tests PASS at r43478 (pre/post-config, corevm/coretest, test, fulltest) - Ubuntu 9.10 amd64 (g++ with --optimize)
14:58 plobsing joined #parrot
15:12 iblechbot joined #parrot
15:24 bubaflub joined #parrot
15:45 fperrad_ joined #parrot
15:46 payload joined #parrot
15:46 whiteknight joined #parrot
15:49 Psyche^ joined #parrot
15:53 whiteknight yay! I've actually been able to post a #ps report today!
15:53 whiteknight first time in months
16:20 darbelo joined #parrot
16:44 hejki left #parrot
16:58 theory joined #parrot
17:09 Coke seen chromatic?
17:09 purl chromatic was last seen on #parrot 8 hours, 47 minutes and 32 seconds ago, saying: We're enlightened.
17:09 * Coke wonders if we could convince purl to also listen on freenode. =-)
17:09 Coke (not say anything, just listen.)
17:15 Zak joined #parrot
17:20 Util Coke: From what Hachi has told me, the main limit on Purl is the IRC network's limit on number of channels that a single nick can simultaneously join. So, *probably*, but I don't know if multi-network code would need to be written.
17:25 whiteknight joined #parrot
17:36 Tene wagle: I hear you wanted to talk to me about scheme on parrot.
17:47 dalek parrot: r43482 | darbelo++ | trunk/src/io/buffer.c:
17:47 dalek parrot: Poking into STRING struct members is bad, but caching pointers into them is worse, since they go stale when the pool gets compacted.
17:47 dalek parrot: This fixes the t/pmc/io.t on OpenBSD
17:47 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43482/
17:54 Zak joined #parrot
18:06 dukeleto 'ello
18:07 whiteknight hello dukeleto
18:11 Tene http://www.research.ibm.co​m/people/d/dfb/papers.html -- several interesting GC papers.
18:12 dukeleto whiteknight: howdy!
18:12 dukeleto Tene: very interesting
18:12 Tene apparently interesting, I should say.
18:12 Tene I haven't read anything there yet.
18:13 dukeleto Tene: i was talking to ward cunningham and he was talking about how newer refcounting algorithms are being found that outperform other types. this looks like an example
18:13 whiteknight_ joined #parrot
18:13 dukeleto ooooooh, it has real-time GC stuff as well
18:15 wagle lemme look at 2.0 later today, and send questions your direction
18:15 wagle Tene: ^^^
18:16 Tene wagle: Great.  I should be around until maybe 4:00 PST, and then gone for the evening hopefully.
18:16 Coke it doesn't surprise me that refcounting is faster.
18:16 wagle Tene: i'm in portland oregon
18:16 Coke I think the point of not refcounting was to avoid bad code.
18:17 Coke Tene: WORK ON PARROT! DOOOOO IT
18:17 dukeleto wagle: you can start hacking with parrot 1.9.0, but 2.0 is a stable API that you should target
18:17 chromatic joined #parrot
18:17 dukeleto wagle: but 2.0 should be out within a few hours or so
18:17 Coke (that said, 2.0 is probably not substantially different than 1.9)
18:17 dukeleto speak of the devil...
18:17 purl Chuck?
18:17 Coke (2.1 is probably going to be more different then 2.0 than 2.0 is from 1.9)
18:17 dukeleto purl, die in a fire
18:17 purl HALP
18:18 Tene wagle: I'm in Salt Lake City, Utah
18:18 Coke Tene: Hey, do you know Tyson Domer? =-)
18:18 Tene Coke: No, I don't.  If you give me an address, I can put a brick through his window, though.
18:18 Coke ROTFL.
18:19 dukeleto #ps is in ~12 min ?
18:19 chromatic yes.
18:19 chromatic NEWS commits are welcome.
18:21 Tene I think that my only commit in the past month was last night, chomping off an extra newline from icu-config that was breaking my generated makefiles.
18:21 Coke tene, that broke my build.
18:21 Tene Coke: wtf?  really?  *sigh*
18:21 Coke (no, just kidding. =-)
18:21 Tene ARGH
18:22 * Coke runs away.
18:22 Tene Yeah, looks like my last commit before that was 1 month + 1 day before.
18:22 Coke it was funny on this cost. =-)
18:22 Coke *coast.
18:22 Tene Coke: remind me to hide snakes in your luggage if I ever run into you at a perl con.
18:23 Coke Absolutely.
18:23 purl Rather.
18:23 Coke Dan?
18:23 purl mailto:dan@sidhe.org or a VMS guy or Eeeeeevil! or late in his parrot child support payments. or fabled to be Parrot's chief architect or MIA or http://www.sidhe.org/~dan/​blog/archives/000435.html or "bailed on Parrot entirely when it became clear that Leo was pushing to get a lot of his bad ideas into the core"
18:23 Coke whoops, wrong dan.
18:24 Coke that's an interesting quote there at the end.
18:24 Tene Hmm.  I never got around to porting squaak to nqp-rx.
18:24 Tene That sounds like a great immediately-before-the-release commit.
18:25 Coke damn. at this point I'll be happy to get one_make merged before 2.1
18:25 Coke s/point/rate/
18:25 Tene I thought that was pretty much done, no?
18:26 Coke lots of chunks left.
18:27 Coke in the middle of killing a config step: can't stop until I'm done.
18:27 Coke (then I can merge back and branch again if I need)
18:29 * dukeleto has another meatspace meeting at the exact time of #ps every week :(
18:29 Tene every week ever?!
18:29 Tene It's okay.  I'll be you for #ps.
18:32 barney joined #parrot
18:33 chromatic #ps time
18:45 Util Coke, kid51: I have added to TT#1393. Equivalent problem reproducable on Darwin/Intel; not covered up in current revision.
18:48 Topic for #parrotis now Parrot 1.9.0 "Blue-fronted Amazon" released! | http://parrot.org | Priorities: merge tt_389_fix and one_make branches | Roadmap: http://icanhaz.com/parrotroadmap | Latest modified TT's: http://icanhaz.com/parrotbugs
18:48 Coke fulltest passed on <some random linux box I have access to>
18:49 chromatic Running fulltest without an optimized build now.
18:49 chromatic msg bacek What's the status of the OrderedHash revamp branch?
18:49 purl Message for bacek stored.
18:52 dalek parrot: r43483 | chromatic++ | trunk/DEPRECATED.pod:
18:52 dalek parrot: [docs] Added two deprecation entries.
18:52 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43483/
18:52 chromatic Tene: see the pge_no_namespace_methods branch.  pmichaud said the right approach is to get the protoobject and fetch the method from there.
18:54 Tene chromatic: is that a branch from trunk or from tt389_fix ?
18:55 chromatic It's a branch from trunk.
18:55 chromatic I added a poison pill to namespace fetching (and we won't commit that chunk).
18:55 Tene Ah.
18:56 chromatic The idea is to fix PGE/TGE in that branch not to fetch methods from the namespace by throwing an exception when they do.  That's much easier to diagnose than failures at a distance.
18:57 Tene That should be straightforward to work on.
18:57 chromatic Does anyone get failures for t/manifest/02-regenerate_file.t, t/manifest/01-basic.t, and t/manifest/03-regenerate_skip.t?
18:57 chromatic Is that a problem with the tests and git-svn?
18:57 Tene I've got $job and then busy all evening, but I can probably work on it tomorrow evening.
18:57 Tene chromatic: I'm running fulltest from git-svn right now
18:59 Util chromatic: those 3 manifest tests pass for me with a fresh checkout of r43482 on Darwin/Intel 10.5
19:00 chromatic Interesting.
19:00 chromatic Util, with svn or git-svn?
19:00 Util chromatic: just svn
19:01 chromatic Good to know, thanks.
19:01 purl i think Good to know, thanks. is there a free version of that I can install (linux or win32) to play with?
19:02 chromatic purl, forget Good to know, thanks.
19:02 purl de nada chromatic
19:03 Util heh. Purl parse FAIL
19:04 chromatic Parrot::Manifest knows nothing of git; it defaults to svk with no .svn directory.  That's the source of the problem.
19:04 chromatic Optimized build... go!
19:05 dukeleto ignore fulltest from a git-svn checkout
19:05 Zak joined #parrot
19:05 dukeleto it is fubared
19:05 dukeleto only trust "make fulltest" from a svn checkout
19:13 chromatic Those are the only failures I see.  Any other concerns?
19:15 * dukeleto does not see anything on fire
19:15 * darbelo sets trunk on fire.
19:18 dukeleto have we tagged and bagged 2.0 yet?
19:19 chromatic Not yet.  I'm updating NEWS.
19:25 dukeleto chromatic++
19:27 chromatic Please review r43484
19:28 Tene hmm.  fulltest has hung for me.
19:28 Tene t/manifest/01-basic.t .................... 1/13
19:28 Tene been there for a long time.
19:28 chromatic Do you have svk installed but unconfigured?
19:28 Tene Yes.
19:29 chromatic Kill svk.
19:30 Coke chromatic: what is the difference between minor and modest?
19:30 dukeleto KILL KILL KILL SVK
19:30 chromatic I have met few modest teenagers.
19:31 Coke can probably delete "realized deprecations" if we have none.
19:31 chromatic I don't know if we do.
19:32 Coke well, if no one adds any to NEWWS before you commit...
19:32 Coke I don't think 'removed typos" is worth an entry in NEWS.
19:32 Coke Also, PIRC is NOT built by default.
19:32 Coke we had to disable it because it breaks teh build.
19:32 chromatic Stupid C++.
19:33 chromatic Okay... I'll give this 15 minutes for review then commit those modifications.
19:33 Coke everything else seems ok.
19:33 Coke wonder if our 2.0 NEWS should point to our last 5 1.x newses.
19:33 cotto_work looks good
19:34 Tene C++ looked like an interesting research project.  Shame it never went anywhere.
19:34 dukeleto lulz
19:39 Coke You can also remove the 'dep checker' as I think that's covered by "makefile improvements"
19:39 AndyA joined #parrot
19:40 chromatic It's a standalone tool.
19:40 Coke your call.
19:41 dalek parrot: r43484 | chromatic++ | trunk/NEWS:
19:41 dalek parrot: Updated NEWS for 2.0 release.
19:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43484/
20:10 joeri joined #parrot
20:19 bacek joined #parrot
20:30 Util chromatic: `make test` has a single failure for me. See http://trac.parrot.org/par​rot/ticket/1401#comment:6
20:34 chromatic That looks bad.
20:38 chromatic Can you get a backtrace, or maybe figure out which pointer is unaligned?
20:38 Util chromatic: I need to get back on the road soon. Is their anything I can do to help quickly?
20:38 Util Oh, OK, let me see
20:41 chromatic Can anyone else reproduce this on Darwin/Intel 10.5?
20:42 GeJ Good morning everyone!
20:45 bacek Good morning
20:46 bluescreen joined #parrot
20:46 bacek chromatic, ordered hash revamp branch isn't mergeable. But it has better implementation of OrderedHash than trunk.
20:47 chromatic How much work remains?
20:48 bacek chromatic, remove tests related on broken behaviour of deleting keys from hash, afair.
20:48 bacek but it was 6 months ago, so I can be wrong.
20:50 chromatic Doesn't sound too difficult.
20:51 bacek chromatic, indeed. I just missed previous deprecation point.
20:51 bacek when "The Release"?
20:52 chromatic I'm waiting to hear back from Util on TT #1401.
20:52 snl joined #parrot
20:52 bacek chromatic, ok.
20:53 chromatic ... or anyone who has access to Intel/Darwin 10.5
20:56 bacek chromatic, I will have access. In couple of hours.
20:56 cognominal joined #parrot
20:56 bacek Unfortunately it will be at $dayjob and I can't spend much time on it...
20:57 chromatic I can do everything else for the release before then.
20:57 bubaflub chromatic: i'm on Intel/Darwin 10.5
20:57 bubaflub do you want me to do a clean build from trunk / make test?
20:58 Util chromatic: backtrace added - http://trac.parrot.org/par​rot/ticket/1401#comment:8
20:59 chromatic bubaflub, can you do a clean build and check Util's backtrace?
21:00 bubaflub chromatic: sure.  clean build coming out now.
21:00 chromatic Thanks!
21:00 bubaflub np.
21:00 chromatic That's an odd string to free.  Hm.
21:03 bubaflub chromatic: i can't reproduce it
21:03 bubaflub either by invoking perl t/harness or by running parrot directly
21:03 chromatic PCRE version?
21:04 bubaflub you'll have to excuse my ignorance - how can i check that?
21:04 bubaflub i don't think i have anything from MacPorts
21:04 bubaflub but lemme double check
21:04 bubaflub wait, i do have one from mac ports
21:04 bubaflub pcre @7.9_0 (active)
21:05 chromatic Util, do you know your PCRE version?
21:06 Util chromatic: how to check?
21:06 payload joined #parrot
21:06 chromatic Good question. bubaflub?
21:07 bubaflub Util: if you have pcre from MacPorts, you can check by typing `port installed`
21:07 bubaflub should dump out a list of installed macports
21:07 bubaflub if not, uh...
21:07 bubaflub hmmm.
21:11 Util I believe that my PCRE was installed by Fink (I don't use MacPorts) back on 10.4, and not updated for 10.5. That is probably the source of the problem.
21:12 chromatic How certain are you?
21:14 Util As certain as I have time to pursue now. The problem is so fragile, that the malloc warning will disappear (though the test failure remains) just because the length of the .pir filename changes. I say Ship It!
21:15 snl joined #parrot
21:15 chromatic That does sound like a version mismatch.
21:19 GeJ `pcre-config --version` should give you what version of pcre you have.
21:19 bubaflub GeJ++
21:22 smash joined #parrot
21:22 smash hello everyone
21:22 Util `/sw/bin/pcre-config --version`      says `4.5`
21:22 Util `/usr/local/bin/pcre-config --version`      says `6.7`
21:22 Util `which pcre-config`                         says `/sw/bin/pcre-config`
21:22 * Util kicks self.
21:23 Util chromatic++, bubaflub++, GeJ++: Thanks!
21:23 * Util starts driving back home.
21:23 darbelo Ship it!
21:30 bubaflub ping treed
21:30 treed ?
21:30 bubaflub are you the person to talk about hacking cardinal?
21:31 allison joined #parrot
21:31 treed I am, yeah.
21:31 plobsing joined #parrot
21:31 bubaflub awesome.  i'd love to start hacking on it.  where should i get started?
21:33 treed The usual thing is to add methods to the built-in classes.
21:33 treed We'd been working on Array.
21:33 smash joined #parrot
21:33 treed I was trying to fix the Object Model to actually match Ruby's but got stuck on it and haven't been able to get moving again.
21:35 bubaflub hmmm... i get an error when i try parrot setup.pir
21:35 Tene what error?
21:35 purl i guess error is KABOOM!
21:36 bubaflub http://gist.github.com/281324
21:36 bubaflub i have my parrot in parrot/ and cardinal in cardinal/
21:36 bubaflub parrot is not installed on my system
21:36 bubaflub in the sense that it's not in my path
21:37 Tene bubaflub: Cardinal builds from an installed parrot.  Install parrot to like $HOME/parrot and add that to your path
21:37 bubaflub Tene: okey dokey.
21:43 GeJ "All tests successful."
21:45 darbelo GeJ: platform?
21:46 GeJ FreeBSD 7.2 amd64
21:46 snl joined #parrot
21:46 bacek joined #parrot
21:48 GeJ smoke report sent
21:49 GeJ 31786
21:50 GeJ will make a request to update the port for FreeBSD once the tarbal is available.
21:50 darbelo GeJ++
21:53 GeJ I need to figure out stuff in the sonfig step first.
21:53 GeJ amn you morning figners!@#!@#
21:54 cotto_work They're out in full force, I see.
21:54 GeJ A clear sign I need more coffee.
21:55 * Coke ponders getting a decaf.
21:55 darbelo Sacrilege!
21:56 Tene I'm pretty restrictive.  I only use caffeine when I notice a particular need for it.  Maybe once a week at most.
21:57 Coke darbelo: it's 5pm here. I don't like being up until 3am anymore. =-)
21:58 darbelo Coke: It's almost 7 pm here, my chances of going to sleep before 3am look slim ;)
21:59 darbelo Which reminds me, I need to up my caffeine IV ;)
22:00 snl joined #parrot
22:02 * darbelo sets his circuit simulator on fire.
22:04 davidfetter o/` come on baby light my circuit simulator o/`
22:05 davidfetter hrm. that doesn't scan very well. sorry, jim.
22:06 darbelo I have only one burning desire,
22:06 darbelo Let me stand next to your circuit simulator
22:06 GeJ hum... is there a known problem in building parrot as root?
22:06 darbelo GeJ: How big a problem?
22:07 Coke ISTR there was an issue with perldoc.
22:08 GeJ dunno exactly. Current version in FreeBSD ports (1.4.0) forbids to build as root... invoking a perldoc issue
22:08 darbelo I think perldoc drops privileges to nobody when run as root, right.
22:08 * Coke 's memory ftw.
22:08 darbelo That causes it to not be able to read files that the rest of the build generated.
22:11 GeJ currently building local checkout as root. I'll see how it goes.
22:21 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31787), fulltest) at r43484 - Ubuntu 9.10 amd64 (g++ with --optimize)
22:23 Tene why would you ever want to build as root?
22:23 Tene That's crazy.
22:24 cognominal joined #parrot
22:26 GeJ Tene: FreeBSD ports system defaults to that behavior.
22:27 Tene I have a personal grudge against FreeBSD anyway, so I'm going to maintain my earlier response. ;)
22:28 darbelo Tene: *personal* grudge?
22:28 plobsing joined #parrot
22:28 Tene darbelo: I had to maintain a horrible FreeBSD server once in the past.  I'm sure it was atypical, but it annoyed me.
22:29 * darbelo facepalms
22:29 darbelo english reading: FAIL
22:30 Tene how did you read it?
22:30 darbelo Actually, english *understanding* FAIL.
22:31 darbelo misunderstood "personal" as "directed to a person"
22:31 Tene Ah.
22:49 kid51 joined #parrot
22:52 cconstantine joined #parrot
23:03 hercynium joined #parrot
23:05 eiro joined #parrot
23:10 * Coke hates dynamic languages that require predeclaring all variables and hash keys.
23:12 tewk_ Coke, what language?
23:15 Coke ColdFusion
23:15 purl ColdFusion is nice... but NT centric... or at http://www.coldfusion.com/ or <CFTAG TYPE=INFESTED> or coral in a previous life
23:15 Coke no, coldfusion is http://www.coldfusion.com/
23:15 purl okay, Coke.
23:16 Coke I want to do some subtotaling, but everything you touch has to be init'd , so you can just do the equiv of $total{$section}{$key}++;
23:16 Coke *can't
23:17 cotto_work That hardly counts as dynamic.
23:25 Tene write a macro!
23:30 Coke I wrote a function.
23:30 Coke CF doesn't have macros, that I'm aware of. =-)
23:33 chromatic Write a compiler in PHP for CF which does have macros.
23:34 cotto_work The release must be making you crabby.
23:34 Coke ... or I could be done already.
23:36 chromatic Updating the PBCs was trickier than I wanted.
23:38 chromatic Did I mention I like git rebase -i?
23:43 Coke git rbi
23:46 ruoso joined #parrot
23:54 chromatic tools/util/crow.pir seems to want compilers/data_json/data_json.pbc in runtime/parrot/languages/data_json/data_json.pbc
23:55 chromatic Even so, it still doesn't work correctly.
23:55 cotto_work The problem with bugs in that script is that it only annoys a single person once per month.
23:56 chromatic I figured someone who likes mangling Makefiles more than I do can fix the first bug.  I might get the second.
23:58 chromatic Ah, a buggy JSON file, thanks to a copy and paste.
23:58 chromatic Did I mention git rebase -i?

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

Parrot | source cross referenced