Camelia, the Perl 6 bug

IRC log for #parrot, 2010-02-05

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 preflex joined #parrot
00:17 Whiteknight joined #parrot
00:29 dalek tracwiki: v25 | kurahaupo++ | ArrayTasklist
00:29 dalek tracwiki: Add "invariant array" section. Remove deprecation of "Array" as it's been done
00:29 dalek tracwiki: http://trac.parrot.org/parrot/wiki/Arr​ayTasklist?version=25&action=diff
00:34 * kurahaupo1 thinks "it's time to learn git, having used SVN all this time". Okay, "apt-get install git". Aaarrgh! "installing gnuit" (GNU Interactive Tools)
00:41 Whiteknight anybody around here know anything about setup.pir and distutils?
00:42 cotto_work definitely fperrad's baby
00:43 cotto_work nobody else seems to have touched distutils
00:43 cotto_work apart from some pod fixes by mikehh
00:44 cotto_work dangerous bus number there
00:46 Whiteknight I'm sort of figuring out my problems anyway
00:48 abqar joined #parrot
00:49 darbelo cotto_work: It also JFW enough that I've never had to touch it.
00:49 cotto_work That's a highly valuable property.
00:50 darbelo It'd get a lot more atention if it sucked.
00:51 he joined #parrot
00:55 Whiteknight okay, well all the new PMCs build, so now it's time to write tests!
01:39 kid51 joined #parrot
02:03 kurahaupo1 joined #parrot
02:22 wknight8111 joined #parrot
02:26 nopaste "kurahaupo" at 118.92.198.223 pasted "ping WhiteKnight++" (25 lines) at http://nopaste.snit.ch/19479
02:26 eternaleye joined #parrot
02:29 JimmyZ joined #parrot
02:30 dukeleto howdy
02:30 purl hello, dukeleto.
02:31 adu left #parrot
02:33 dalek parrot: r43720 | jkeenan++ | trunk/t/tools/pmc2cutils (3 files):
02:33 dalek parrot: Use resizablepmcarray.pmc instead of the now departed array.pmc.
02:33 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43720/
02:34 JimmyZ_ joined #parrot
03:23 dalek parrot: r43721 | jkeenan++ | trunk (85 files):
03:23 dalek parrot: Merge tt1393_retcon branch into trunk.  Suggested by bacek++ and lithos++.
03:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43721/
03:23 dalek parrot: r43722 | jkeenan++ | branches/tt1393_retcon:
03:23 dalek parrot: Branch has been merged into trunk and is no longer needed at HEAD.
03:23 purl i already had it that way, dalek.
03:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43722/
03:23 dalek parrot: r43723 | jkeenan++ | tags/tt1393_retcon-43651:
03:23 dalek parrot: Branch to which tag corresponded has been merged into trunk; tag is no longer needed.
03:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43723/
03:31 * kid51 must sleep
03:31 purl $kid51->sleep(8 * 3600);
03:42 dalek TT #1420 closed by jkeenan++: Failed assertion when running 240th test in 01-regex.t in GGE
04:04 * plobsing reads config/gen/makefiles/root.in and is very confused by parrot_config.o
04:05 abqar joined #parrot
04:06 plobsing if parrot_config.o isn't part of libparrot, don't all embedders have to provide an equivalent?
04:08 dukeleto plobsing: good question
04:08 purl Yeah, it is. I'm stumped.
04:09 kurahaupo1 joined #parrot
04:10 plobsing the problem as I see it is that libparrot is built *before* miniparrot
04:10 plobsing which is getting in my way trying to massage nativecall.pir into the build
04:11 plobsing if miniparrot is required for building stuff, libparrot shouldn't be around yet
04:16 dukeleto plobsing: that seems reasonable
04:17 chromatic There's little point to miniparrot then; everything it needs is in libparrot, really.
04:18 plobsing I'm looking to take full nci out of miniparrot, leaving a stripped-down, required-for-core-only nci
04:19 plobsing which is a mostly static thing
04:21 dukeleto do we need a minilibparrot?
04:21 dukeleto or libminiparrot, rather
04:21 plobsing dukeleto: then we could use miniminiparrot to make that :p
04:22 plobsing I was thinking of just linking against $(O_FILES)
04:52 mikehh joined #parrot
04:53 chromatic A minimal NCI would be okay.
04:53 chromatic It helps Lorito.
04:54 plobsing is it possible to add to a static library (portably) ?
04:55 plobsing s/add/append/
04:58 chromatic I don't know of one.
04:58 chromatic I don't know of (a portable way).
05:00 dukeleto here is the CPU that is in the BUG http://www.freescale.com/webapp/sps​/site/prod_summary.jsp?code=i.MX31 (Freescale i.MX31 ARM)
05:02 kurahaupo "ar -a archive new-member"
05:05 kurahaupo Assuming you have ".a" static libs?
05:05 chromatic Sure, on a modern Unix-like system.
05:05 chromatic I'm not sure we can rely on that for Mac OS X.
05:06 kurahaupo That's BSD under the hood isn't it?
05:06 chromatic More like NeXT.
05:06 Casan joined #parrot
05:06 chromatic In a very technical sense, it's *a* BSD, but it doesn't work like, say, FreeBSD 7.
05:07 * kurahaupo curses tiny on-screen keyboard and his clumsy thumbs. Txting is sooo slow. :-(
05:08 kurahaupo Anyone have a shell on a Mac?
05:11 kurahaupo "file /bin/sh"
05:12 kurahaupo "file /lib/libc.a"
05:13 nopaste joined #parrot
05:13 kurahaupo If those work the ar should too.
05:16 chromatic I don't think they use ar to make static libraries though.
05:17 treed [21:17] [inara:/Users/treed]% file /lib/libc.a
05:17 treed /lib/libc.a: cannot open `/lib/libc.a' (No such file or directory)
05:18 kurahaupo It's usually wrapped
05:18 nopaste joined #parrot
05:18 TonyC joined #parrot
05:29 plobsing_ joined #parrot
05:50 dalek parrot: r43724 | plobsing++ | branches/nativecall_pir_build:
05:50 dalek parrot: create branch to rearrange build to use nativecall.pir
05:50 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43724/
06:06 dalek parrot: r43725 | plobsing++ | branches/nativecall_pir_build/​config/gen/makefiles/root.in:
06:06 dalek parrot: break dependancy of miniparrot on libparrot
06:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43725/
06:15 eternaleye (backlogging) (re RPA imperfections) What about Judy trees?
06:19 dukeleto wow, i am reading about Jazelle - http://en.wikipedia.org/wiki/Jazelle . ARM cpus with jazelle can execute java bytecode in hardware
06:22 dukeleto The published specifications are very incomplete, being only sufficient for writing operating system code that can support a JVM that uses Jazelle ... his gives ARM Holdings considerable control over which JVMs are able to exploit Jazelle. It also prevents open source JVMs from using Jazelle.
06:22 dukeleto s/his/this/
06:25 eternaleye dukeleto: Follow the link to the ARM page and go to the ThumbEE entry, it addresses that
06:27 dukeleto eternaleye: yes, i was reading that. thankfully thumbEE is replacing jazelle
06:31 eternaleye Indeed.
06:34 dukeleto so it looks like it is very similar to the iphone ARM cpu. porting to BUG may give us the iphone for free
06:34 eternaleye Yeah, but it'll ony be available to jailbroken ones since Apple has decreed 'no interpreters'
06:34 eternaleye *only
06:35 dukeleto eternaleye: yes, of course.
06:35 purl Indubitably.
06:35 davidfetter joined #parrot
06:36 dukeleto iphone is a ARM1176JZ(F)-S and the BUG has a ARM1136J(F)-S. kissing cousins
06:36 eternaleye I still can't wrap my head around the kind of thinking neccessary to make a decision like that
06:36 szbalint reminds me now that im at the airport i should install parrot
06:37 szbalint with 2 hours to kill it should be ok
06:37 dukeleto szbalint: installing parrot should be pretty quick
06:38 davidfetter szbalint, where to?
06:38 szbalint fosdem
06:38 szbalint same as you
06:38 dukeleto davidfetter: both of you will be at fosdem
06:38 davidfetter szbalint, w00t! i'm in .be already, getting over jetlag
06:38 szbalint cool
06:38 davidfetter any chance you'll stop by the pg booth?
06:39 szbalint btw im with the pg group
06:39 szbalint andreas scherbaum and bunch
06:39 davidfetter dukeleto, szbalint, i'd like to coordinate with you on a lightning talk if possible
06:39 davidfetter oh, sweet. i didn't know you knew each other :)
06:39 szbalint are you at the pg hotel?
06:39 davidfetter yep
06:40 davidfetter well, i will be today
06:40 szbalint sweet
06:40 dukeleto davidfetter: spill the beans
06:40 purl You'll never get me to rat them out!
06:40 * davidfetter visiting relative in ghent
06:40 szbalint i will arrive around 1pm
06:40 davidfetter dukeleto, well, i was thinking about a (brief) live demo with what we've got to show it's not vapor
06:40 davidfetter what can we show?
06:41 szbalint dukeleto: parrot on my phone that is. sorry irssi is a bit cramped on this smallish screen :)
06:42 chromatic Ask Tene for a language interoperability demo.
06:42 davidfetter hrm
06:43 Tene language interop is hosed right now.
06:43 davidfetter good to know
06:43 davidfetter is it in 2.0.0?
06:43 Tene I might be able to fix it up this weekend.
06:43 Tene davidfetter: the problem is that none of the languages other than rakudo have been updated to match the new language interop API after pmichaud changed it.
06:44 davidfetter i confess i've been working with packaged parrot + bleeding-edge pg to get things working. will have to spin up on bleeding-edge parrot, too
06:44 davidfetter Tene, ok, are there people whose chains should be getting rattled on this?
06:45 dukeleto davidfetter: what do you want to show off?
06:45 Tene davidfetter: mine.  I've been the major HLL interop maintainer in the past.
06:45 dukeleto davidfetter: you want to do a lightning talk about PL/Parrot?
06:45 Tene Life has just been interfering lately.
06:47 kurahaupo joined #parrot
06:48 davidfetter Tene, how much would parrot people need to know to be able to do changes on the other HLLs?
06:48 davidfetter dukeleto, yep
06:49 Tene davidfetter: not much.  Just need to read the HLL interop spec in docs/pdds/something.pod and then look at a couple of pieces of rakudo and nqp for examples.
06:49 davidfetter hrm
06:51 dukeleto davidfetter: what are you going to demo? spell out an example of a decent small live demo.
06:53 davidfetter i'm thinking a step-through of compiling it & running the regression tests
06:53 davidfetter + Q&A
06:53 davidfetter it'a amazing what you can do in 5 minutes :)
06:54 dukeleto davidfetter: sounds feasible
07:03 dukeleto darbelo: ping! perl -V on BUG: http://gist.github.com/295587
07:22 kurahaupo1 treed: libc.a is probably in /usr/lib, but might be /System/lib. Try something like "f=$(locate libc.a) && ( file $f ; ar tv $f ) | more"
07:23 dukeleto msg darbelo check this out: http://pastebin.com/m4422a776
07:23 purl Message for darbelo stored.
07:24 dukeleto kurahaupo1, treed: /usr/lib/libc.dylib
07:24 kurahaupo1 That doesn't sound like a static library. :-(
07:24 kurahaupo1 What does 'file' say about it?
07:25 dukeleto kurahaupo1: Mach-O dynamically linked shared library
07:25 kurahaupo1 dukeleto: echo /usr/lib/*.a | tell kurahaupo ?
07:31 dukeleto kurahaupo: http://gist.github.com/raw/295605/a5556​3cc992e267a95c3c2386589d38b39b039dd/*.a
07:32 dukeleto kurahaupo: /usr/lib/libstdc++-static.a ?
07:32 kurahaupo dukeleto: "ar tv /usr/lib/libstdc++-static.a | more"
07:33 kurahaupo Look for anything that's not "*.o", or has weird names like "__INDEX__.o"
07:34 dukeleto kurahaupo: ar: /usr/lib/libstdc++-static.a is a fat file (use libtool(1) or lipo(1) and ar(1) on it)
07:34 dukeleto ar: /usr/lib/libstdc++-static.a: Inappropriate file type or format
07:34 kurahaupo Okay, I'm officially out of my depth on MacOS. :-(
07:34 kurahaupo (What's a "Fat file"?)
07:36 dukeleto kurahaupo: ENOCLUE
07:37 szbalint i think its a dual binary or somethin
07:37 szbalint g
07:40 szbalint see wikipedia  on universal binary
07:47 dukeleto szbalint: it has 4 architectures in it
07:48 dukeleto szbalint: http://gist.github.com/295617
07:49 szbalint pretty interedting :)
07:50 szbalint ok gotta go, boarding.
07:50 dukeleto szbalint: that is a word ;)
07:50 dukeleto szbalint: have fun!
07:50 purl Don't do anything I wouldn't do!!!
07:50 szbalint yesh im fat fingering on this tiny keyboard
07:51 szbalint bbl.
07:53 mikehh joined #parrot
07:55 dukeleto mikehh: good localtime()
07:55 kurahaupo joined #parrot
08:06 mikehh joined #parrot
08:16 iblechbot joined #parrot
08:46 fperrad joined #parrot
09:05 bacek joined #parrot
09:06 bacek o hai
09:33 payload joined #parrot
09:36 payload joined #parrot
09:37 mikehh joined #parrot
10:14 payload joined #parrot
10:25 davidfetter joined #parrot
11:05 clinton joined #parrot
11:46 riffraff joined #parrot
12:08 ruoso joined #parrot
12:08 szbalint davidfetter: ping. arrived at hotel
12:09 davidfetter szbalint, i'm in room 54
12:09 davidfetter (5th floor)
12:09 davidfetter hungry?
12:09 purl hungry is, like, "And thou shalt eat the fruit of thine own body, and the flesh of thy sons and of thy daughters, which the Lord thy God hath given thee." Deuteronomy 28:53 or (see the first half of the verse) or http://badgas.co.uk/moments/moment_089.jpg
12:09 szbalint yep
12:09 davidfetter how long do you need to settle in?
12:10 davidfetter i was thinking about getting some .be food over at le chou de bruxelles
12:10 szbalint im settled. shall we meet in the lobby?
12:10 davidfetter sure :)
12:10 davidfetter lemme grab a jacket & phone
12:10 szbalint 41 is my room btw
12:11 szbalint ok. omw then.
12:13 payload joined #parrot
12:19 bluescreen joined #parrot
13:32 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32067), fulltest) at r43725 - Ubuntu 9.10 amd64 (gcc with --optimize)
13:40 tetragon joined #parrot
13:55 cognominal joined #parrot
13:56 payload joined #parrot
13:58 particle1 joined #parrot
14:31 particle1 joined #parrot
14:47 dalek parrot: r43726 | mikehh++ | trunk/config/auto/icu.pm:
14:47 dalek parrot: move $icushared substitution into if to avoid warning message if not defined or empty
14:47 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43726/
14:53 payload joined #parrot
15:06 plobsing joined #parrot
15:17 payload joined #parrot
15:29 Psyche^ joined #parrot
15:56 bubaflub joined #parrot
16:08 mikehh joined #parrot
16:12 fperrad_ joined #parrot
16:22 whiteknight joined #parrot
16:23 bubaflub joined #parrot
16:26 iblechbot joined #parrot
16:29 whiteknight good mornig parrot
16:35 theory joined #parrot
16:40 particle that's a matter of opinion
16:46 davidfetter joined #parrot
16:59 davidfetter what's the svn revision # for parrot 2.0.0?
17:06 szbalint nm, downloaded it from parrot.org :)
17:08 Coke davidfetter: don't assume that a specific trunk corresponds to a release.
17:09 davidfetter Coke, ok
17:09 Coke if you're pulling it from the repo, go for tags/RELEASE_2_0_0
17:09 Coke it happens to work that way most of the time. but don't rely on it. =-)
17:16 szbalint does make install work for parrot 2.0 ?
17:17 szbalint (as in, is it recommended?)
17:17 Coke ... since 1.0, surely.
17:17 Coke when working on partcl, for example, I ONLY work against an installed parrot.
17:17 szbalint oh ok. I had outdated information then :)
17:18 Coke and, 'make install' installs everything a language developer would need.
17:18 Coke (used to have to jump through hoops for that, now you have to jump through hoops to not.)
17:21 bubaflub joined #parrot
17:31 jan joined #parrot
17:52 cognominal joined #parrot
18:03 ash_ joined #parrot
18:05 cotto_w0rk joined #parrot
18:16 cotto_work good mooring
18:16 whiteknight urg, pmc2c is silently ignoring all the VTABLEs and METHODs in my PMC, and I don't know why
18:16 cotto_work Are your changes on git?
18:17 cotto_work s/git/github/
18:17 cotto_work and which ones are b0rked
18:17 cotto_work ?
18:17 whiteknight http://github.com/Whiteknight/parrot-data-struc​tures/blob/master/src/pmc/resizablepmcqueue.pmc
18:18 whiteknight this PMC, for some reason. The generated C code doesn't have any of the ATTRs, VTABLEs, or METHODs
18:18 cotto_work give me a sec to install parrot
18:19 darbelo whiteknight: doble check your vtable names or for unbalanced comments.
18:20 darbelo Silelntly ingoring you is pmc2c's usual mode of dealing with typos.
18:22 chromatic joined #parrot
18:26 darbelo whiteknight: Found it!
18:26 darbelo pmc2c--
18:26 whiteknight eh?
18:27 darbelo s/void*/void */ in you ATTRS
18:27 whiteknight ah, that did it!
18:28 whiteknight darbelo++
18:28 cotto_work I *know* fixed that at in pmc2c some point.
18:28 cotto_work istr that it got reverted for some reason
18:28 * cotto_work goes off to do something
18:28 tewk parsing with regexs is always brittle
18:34 darbelo Think of it as enforcing a coding standard. The '*' belongs next to the variable name anyway ;)
18:36 cotto_work That failure mode needs to be fixed.  Anyone want to file a ticket?  Bonus points if you mentioned how much time finding it wasted this morning.
18:36 plobsing joined #parrot
18:39 Coke there is already a ticket.
18:40 Coke TT #1041
18:40 cotto_work There's http://trac.parrot.org/parrot/ticket/1041, but that's not closeable.
18:41 Coke sure it is.
18:41 Coke "change pmc2c.pl so it passes through anything it is not changed."
18:42 * darbelo wants a "pmc2c sucks and must die" ticket
18:42 * Coke guesses that it doesn't cover this current complaint, though.
18:43 PerlJam "rewrite pmc2c using nqp-rx"   :-)
18:43 cotto_work That's closeable, if big
18:43 darbelo Thare's even a branch for it already.
18:43 cotto_work yup
18:44 cotto_work speaking of which, does nqp-rx support mmd yet?
18:44 cotto_work istr that that's something that'd make building pmcc with pure nqp easier
18:45 bubaflub dumb question: is pmc2c called late enough in the build process to use nqp?
18:45 darbelo Mu.
18:46 whiteknight bubaflub: no. pmc2c is called early
18:46 whiteknight but we could bootstrap
18:48 bubaflub whiteknight: ok; so a conversion of pmc2c to nqp would require shifting some of the build process around
18:48 plobsing I don't see how that would work. You can't even have miniparrot without pmcs
18:48 whiteknight bubaflub: yes. Would have to include the generated src/pmc/*.c files in the repo, use those to build parrot stage 0, build NQP, call pmc2c-nqp, rebuild parrot stage 1
18:49 darbelo Yes, but only for a few core PMCs, the rest can be made into dynpmcs.
18:49 Coke ... that's a huge huge change.
18:49 darbelo plobsing: check in the generated code.
18:49 plobsing yeah, that requires more than a shuffle in build order
18:50 bubaflub is pmc2c too complicated for pure PIR?
18:50 cotto_work darbelo++
18:50 Coke that doesn't change the amount of bootstrapping required.
18:50 plobsing bubaflub: you're welcome to try ;-)
18:50 Coke requiring pir still means we need parrot to build parrot.
18:50 bubaflub Coke: ah. true.
18:50 whiteknight ever tried to build GCC from source?
18:50 Tene bubaflub: nqp generates PIR, and doesn't have a runtime library, so you just check in the generated PIR.
18:51 whiteknight even a two-stage bootstrap for Parrot is not as bad as GCC's build
18:51 plobsing Tene: did not know that. that might come in handy for me.
18:52 cotto_work It has a runtime library, but it consists of a definition of 'say'.
18:54 plobsing one issue with checking in generated files is that, up to date, I can't get make to accept rules to generate the generated files (circular deps)
18:55 whiteknight plobsing: it's the same way we manage the generated flex/bison files from IMCC
18:55 whiteknight we include the .c files in the repo for people who don't have flex/bison
18:55 plobsing whiteknight: not exactly the same. in this case you have PMC.c files that depend on NQP-rx and Parrot which depend on PMC.c files.
18:56 plobsing flex/bison doesn't have cyclic deps
18:56 whiteknight plobsing: deps are only circular if we make the same files be the inputs as the outputs
18:57 whiteknight we could have a separate src/pmc/bootstrap folder to hold pre-built PMCs that don't get rebuilt by the standard make target
18:57 whiteknight would need to run "make bootstraps" to update them explicitly
18:57 plobsing hmmm... sounds useful, if a little probable to get forgotten once too often
18:58 whiteknight src/pmc/bootstrap/*.pmc > build parrot > build nqp > build pmc2c-nqp > src/pmc/*.pmc
18:59 cotto_w0rk joined #parrot
18:59 plobsing I'm having the same issues I just described with nativecall.pir, so I'll give that a try
19:01 plobsing whiteknight: would it be feasible to have some kind of commit hook to reject commits that de-sync bootstrap with source?
19:01 whiteknight plobsing: if the hook were sufficiently smart that it didn't raise false positives
19:02 cotto_work sounds like an expensive hook
19:02 cotto_work I'd put it in the "Don't Do That" category.
19:04 tewk I'm more in favor of build-bots that yell after the fact than commit hooks that stall parallel development.
19:05 kurahaupo_mobi Like a style-checker?
19:05 plobsing I don't care how we do it, I just don't want to increase the chances of accidentally inconsistent commits
19:06 cotto_work a codingstd test could work
19:06 cotto_work or codetest
19:06 purl somebody said codetest was part of fulltest,  distro_tests is part of fulltest, but not of codetest
19:06 plobsing because so many people run those before they commit
19:07 Tene Well, let's ask how smart it really would need to be.  What are the situations where updating the source files or the generator would not also need to update the generated versions?
19:07 ash_ joined #parrot
19:08 plobsing doc patches, small corner-case bugfixes
19:09 kurahaupo_mobi Seems to me that the prebuilt .c files are part of the build tools, and therefore only have to be "new enough and not broken". Exact sync is NOT desirable.
19:10 dalek tracwiki: v7 | darbelo++ | ItsABughunt
19:10 dalek tracwiki: The main complaints against this code have been adressed, let's remove it now and add it back when new complaints arise.
19:10 dalek tracwiki: http://trac.parrot.org/parrot/wiki/I​tsABughunt?version=7&action=diff
19:11 darbelo I wonder if packfiles/pbc deserve a mention in that page...
19:12 cotto_work yes
19:12 cotto_work yes
19:12 cotto_work also, yes
19:13 darbelo I, however, am not familiar enough with it's problems to add it. And I don't want to be either.
19:20 cotto_work chromatic would be best for that.  I can add a little though.
19:26 Infinoid I still have scars from that code
19:26 dalek tracwiki: v8 | cotto++ | ItsABughunt
19:26 dalek tracwiki: some initial comments on freeze/thaw and packfile
19:26 dalek tracwiki: http://trac.parrot.org/parrot/wiki/I​tsABughunt?version=8&action=diff
19:27 cotto_work pmc2c could go there too
19:31 Limbic_Region joined #parrot
19:39 joeri joined #parrot
19:55 particle Infinoid: how do you like snow?
20:05 whiteknight we're going to get a shittonne around here
20:09 bluescreen joined #parrot
20:11 plobsing joined #parrot
20:13 dalek parrot: r43727 | darbelo++ | trunk/src/pmc/imageio.pmc:
20:13 dalek parrot: Remove a unnecessary #define, this happen on another file now.
20:13 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43727/
20:14 janus joined #parrot
20:15 whiteknight when I type "git push", I get a long warning about how I did not specify any refspecs to push. How do I set it up so it just pushes to github and doesn't give me crap about it?
20:17 darbelo nopaste your .git/config ?
20:17 whiteknight nopaste?
20:17 purl i think nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels)  or http://paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at http://www.extpaste.com/ or http://paste.scsys.co.uk (for #catalyst, #dbix-class, #moose  and others) or http://gist.github.com/ or paste or gtfo or tools/dev/nopaste.pl or https://trac.parrot.org/parrot/br​owser/trunk/tools/dev/nopaste.pl
20:18 nopaste "Whiteknight" at 173.12.37.77 pasted ".git/config for darbelo++" (11 lines) at http://nopaste.snit.ch/19481
20:21 bubaflub whiteknight: under my gitconfig i have a section for push where i have default = current
20:21 bubaflub that should allow you to push whatever branch you are currently on
20:23 darbelo whiteknight: what bubaflub said.
20:23 bubaflub whiteknight: i can nopaste my git config for you to look at, mostly stuff i stole from other places
20:23 darbelo "git config push.default current"
20:23 whiteknight okay, so "git config push.default current"
20:23 whiteknight ok
20:25 Infinoid particle: snow purdy.
20:29 darbelo Hmm.
20:29 darbelo Such a small patch, and so big a fail...
20:30 darbelo http://nopaste.snit.ch/19482
20:31 bubaflub darbelo: yikes.
20:33 cotto_work 'splode
20:34 darbelo Well, it does makes sense that it should happen. We are playing with raw pointers in there. Maybe we should just block the gc unconditionally and delete the comment.
20:34 cotto_work it'd be less mysterious
20:34 plobsing darbelo: where are we playing with raw pointers?
20:35 darbelo We give the Packfile primitives pointer into buffers. Not that we have a choice about it.
20:36 plobsing Can't we pin buffers temporarily accross calls?
20:37 darbelo Right now we're outright blocking the gc. I was experimenting with disabling that.
20:38 bacek joined #parrot
20:38 plobsing darbelo: then how was I getting GC bugs back in the pmc_freeze_cleanup branch?
20:39 darbelo Those were in freeze, we block gc for thaw.
20:39 plobsing fun
20:39 darbelo see http://nopaste.snit.ch/19482 and line 90 of src/pmc_freeze.c
20:44 plobsing I can understand the logic of not wanting GC during an allocation only period, but shouldn't it mark/sweep first?
20:46 darbelo plobsing: Probably, that comment is several years old.
21:12 kurahaupo_mobi Which GC implementations show problems during freeze?
21:13 plobsing kurahaupo_mobi: none are showing ATM, but we ran into issues when we wanted to make STRINGs out of Buffers
21:18 kurahaupo_mobi Mutating pobj to str in-place or new string referring to same mem block?
21:19 plobsing new string, copying mem-block
21:19 plobsing there was no api to do it with a memory managed memblock AFAICT, so we made one
21:20 plobsing Basically, any Parrot function that takes a 'char *' buffer should also provide a 'Buffer *' version
21:21 plobsing because you can't just pass in Buffer.bufstart (it can move)
21:27 plobsing In fact, there's little to nothing you can safely do with Buffer.bufstart. Maybe we should create macros to poke it to provide more of what C programmers expect out of a buffer (eg: get/set byte at index)
21:29 darbelo It looks to me like buffers aren't very 'first class'.
21:30 plobsing I think they should be. That or let me make blatently break the encapsulation around STRING.
21:30 darbelo I get the feeling the 'higher' layers of parrot are better off not using them.
21:31 darbelo Or maybe we should improve the interface to STRING, so that we don't need Buffer.
21:32 plobsing but the whole point of STRING is that its so much more than a buffer
21:32 plobsing and you shouldn't be treating it like on
21:32 plobsing s/on/one/
21:33 ash_ is that the basic difference between an array of chars in C and a std::string in C++? one's a glorified buffer, the other is a string object
21:34 darbelo ash_: In a way.
21:35 dalek parrot: r43728 | plobsing++ | branches/nativecall_pir_build (6 files):
21:35 dalek parrot: remove miniparrot dependancy on src/nci.o using src/bootstrap/nci.c
21:35 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43728/
22:03 Whiteknight joined #parrot
22:04 cotto_work pmichaud, ping
22:09 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32069), fulltest) at r43728 - Ubuntu 9.10 i386 (g++ with --optimize)
22:26 TiMBuS joined #parrot
22:27 darbelo Say, do the keys used in config_lib.pasm need a deprecation notice before removal?
22:29 * darbelo guesses not.
22:30 cotto_work joined #parrot
22:32 cotto_work msg pmichaud nqp-rx has been really quiet lately.  Is this because it's "done" or from lack of tuits?
22:32 purl Message for pmichaud stored.
22:41 japhb cotto_work, not to speak for him, but ... he said a couple days ago that he will be on very odd hours this weekend and in to next week, so he may be slow to reply.  Also, last I heard, he was putting most of his energy into getting Rakudo's ng1 branch landed as the new master; I suspect that NQP-rx work will be delayed at least until after that happens.  In the mean time, you can see/add to the requested features list at http://wiki.github
22:41 japhb .com/perl6/nqp-rx/requests .
22:41 japhb Oh nice, split the URL, sigh.
22:41 japhb http://wiki.github.com/perl6/nqp-rx/requests
22:42 japhb (Of course, I haven't been watching #perl6 this week, ng1 may have landed by now.)
22:49 cotto_work thanks
22:52 darbelo Is there a reason we care about where perl5 keeps it's headers?
22:53 darbelo http://nopaste.snit.ch/19483 removes that from Configure and I see no breakage.
22:53 darbelo I'm also unable to find any use at all fo that configure key.
22:55 cotto_work How old is that chunk of code?
22:56 darbelo let me check.
22:57 japhb darbelo, used by the Perl 5 embedding somehow (Blizkost)?  Though I would have assumed that would be handled by the Blizkost config/make
22:57 cotto_work It's pretty old.  Make sure to run all tests but if nothing fails, rip it out.
22:57 cotto_work It predates blitzkost considerably.
22:57 darbelo lat touche in r24409
22:58 cotto_work http://trac.parrot.org/parrot/brow​ser/trunk/config/init/defaults.pm?​annotate=blame&rev=42577#L190
22:59 darbelo And THAT was code-shuffling, not an addition.
23:00 darbelo Last time before that was r17104.
23:07 jsut_ joined #parrot
23:17 hercynium joined #parrot
23:26 Limbic_Region joined #parrot
23:36 joeri joined #parrot
23:45 dalek parrot: r43729 | darbelo++ | trunk/config/init/defaults.pm:
23:45 dalek parrot: Remove unused configure key.
23:45 dalek parrot: review: http://trac.parrot.org/parrot/changeset/43729/
23:49 kid51 joined #parrot
23:53 patspam joined #parrot

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

Parrot | source cross referenced