Camelia, the Perl 6 bug

IRC log for #parrot, 2009-02-16

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:01 kid51 joined #parrot
00:08 TiMBuS joined #parrot
00:10 NotFound joined #parrot
00:10 NotFound hi
00:33 Theory joined #parrot
00:47 rg coke: i don't think RT#36086 is quite as easy as you think (or my understanding of i386 assember is just too lacking)
02:04 Coke I will be happy if there is a better understanding of a) if the problem is still occurring, and b) what the actual cause is.
02:05 kid51 joined #parrot
02:13 Tene joined #parrot
02:14 rg definitely yes to a)
02:15 rg not sure about the cause, though. i'm still looking
02:32 Andy joined #parrot
03:36 janus joined #parrot
03:58 Andy joined #parrot
04:20 rurban_ joined #parrot
04:22 dalek parrot: r36781 | petdance++ | trunk/src/pmc/integer.pmc:
04:22 dalek parrot: lotsa consting
04:22 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36781/
04:52 bacek joined #parrot
04:53 dalek parrot: r36782 | allison++ | trunk/config/init/install.pm:
04:53 dalek parrot: [config] Fix undefined variable warning in step tests.
04:53 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36782/
05:12 dalek parrot: r36783 | allison++ | trunk/src/library.c:
05:12 dalek parrot: [install] Fix versioned library search paths, generating the paths from the
05:12 dalek parrot: config data.
05:12 purl hmmm... config data is naturally nested and indented even when i write it in perl or apache.
05:12 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36783/
05:24 cotto no stupid bot, config data is <reply>
05:24 cotto no purl, config data is <reply>
05:24 purl okay, cotto.
05:49 dalek parrot: r36784 | allison++ | trunk/src/library.c:
05:49 dalek parrot: [install] Remove the hard-coded install paths for library searching, and only
05:49 dalek parrot: use the paths pulled from config.
05:49 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36784/
06:18 chid joined #parrot
06:55 dalek parrot: r36785 | allison++ | trunk/tools/dev/install_dev_files.pl:
06:56 dalek parrot: [install] Also install nqp.
06:56 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36785/
06:58 dalek parrot: r36786 | allison++ | trunk/MANIFEST.generated:
06:58 dalek parrot: [install] Also install generated .pbc file for nqp.
06:58 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36786/
07:01 namenlos joined #parrot
07:06 dalek parrot: r36787 | allison++ | trunk:
07:06 dalek parrot: [install] Mark 't/' directories as "test" package, so they never get installed
07:06 dalek parrot: with the rest of the language/compiler/tool.
07:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36787/
07:09 dalek parrot: r36788 | allison++ | trunk/MANIFEST.SKIP:
07:09 dalek parrot: [cage] Apparently someone changed the svn:ignore property on languages/c99.
07:09 dalek parrot: Committing change from regenerated manifest files.
07:09 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36788/
07:12 namenlos joined #parrot
07:16 bacek perl6: say (1..2) Z (5..7)
07:16 polyglotbot OUTPUT[1526␤]
07:22 uniejo joined #parrot
07:23 integral joined #parrot
07:42 chid I don't understand how to run perl6 with parrot
07:43 khatar chid, did you compiled perl6.pbc ? ?
07:44 chid no, I'm on win32 vista, just installed the parrot binary
07:44 khatar try to cd langages/perl6
07:44 khatar make
07:44 khatar make spectest
07:44 purl i guess make spectest is failing more than usual too
07:44 chid :/
07:45 chid in the per6 directory there is only one file, perl6.pbc, so I type nmake in that dir?
07:45 TiMBuS you run the pbc with parrot
07:46 chid oh
07:46 chid so, it would be, parrot pbc <file>?
07:47 TiMBuS just parrot <file.pbc> should do
07:47 chid it doesn't work
07:47 purl It's a Y2K error!  Panic!  Sue!
07:47 chid "load_bytecode" couldn't find file 'PCT.pbc'
07:48 TiMBuS hm
07:48 chid current instr.: '' pc 148 (src\classes\Object.pir:20)
07:48 chid perhaps the win32 port is broken?
07:49 TiMBuS i know the svn build should work
07:49 chid If it's not an easy fix, I'll just stay with perl 5
07:49 TiMBuS i assume you got the windows binary package?
07:50 chid TiMBuS yes
07:50 TiMBuS ill see whats in it. it shouldnt need to be built though
07:51 chid thanks
07:52 TiMBuS if parrot is working and it can't find PCT.pbc, it might just need the right PATH setting
07:53 chid perhaps..
07:53 purl somebody said perhaps was some alternative?
07:55 TiMBuS purl, forget perhaps
07:55 purl TiMBuS: I forgot perhaps
07:55 chid lol
07:55 chid I have nothing in my /library/PCT folder except for readme
07:56 TiMBuS hrm
07:56 dalek rakudo: 2f489f2 | (Moritz Lenz)++ | src/pmc/perl6multisub.pmc:
07:56 TiMBuS the setup wont work under wine so i cant see
07:56 dalek rakudo: fix spelling of function name in perl6multisub.pmc
07:56 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​f489f2c8f65a48b6ee9376ac4da9c5fba07d8f1
07:56 shorten dalek's url is at http://xrl.us/begeqg
07:56 dalek rakudo: 2c972d7 | (Moritz Lenz)++ | t/spectest.data:
07:56 dalek rakudo: [t/spectest.data] add two passing regression tests
07:56 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​c972d7513e9cb1912a1641a128e9247ed9ea03a
07:56 shorten dalek's url is at http://xrl.us/begeqi
07:56 TiMBuS but that sounds wrong
07:56 dalek rakudo: 314d8ce | (Moritz Lenz)++ | tools/update_passing_test_data.pl:
07:56 dalek rakudo: handle different build layout in in update_passing_test_data.pl
07:56 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/3​14d8cefe06a592e854fa78605fcd20ad92726fa
07:56 shorten dalek's url is at http://xrl.us/begeqk
07:56 dalek rakudo: a3fd610 | (Moritz Lenz)++ | tools/test_summary.pl:
07:56 dalek rakudo: handle different build layouts in test_summary.pl
07:56 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​3fd6100f41b1364f0ac84ec90da4f4f1020fe6e
07:56 shorten dalek's url is at http://xrl.us/begeqn
07:56 dalek rakudo: 27c64b5 | (Moritz Lenz)++ | tools/autounfudge.pl:
07:56 dalek rakudo: update autounfudge.pl to deal with both build layouts
07:56 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​7c64b578ce40a20d084dc76958f94327a0b803e
07:56 shorten dalek's url is at http://xrl.us/begeqp
07:57 dalek rakudo: a93b241 | (Moritz Lenz)++ | README:
07:57 dalek rakudo: [README] update to reflect current status
07:57 dalek rakudo: also tell people to use git-format-patch if possible.
07:57 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​93b2410b85d0d00bf6727570832ead86956ca30
07:57 shorten dalek's url is at http://xrl.us/begeqr
07:57 dalek rakudo: beac378 | (Moritz Lenz)++ | README:
07:57 dalek rakudo: [README] link to Chris Dolan's "steps to create a patch".
07:57 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/b​eac3787ddec644d79dcb6152abcda1f4ce59a89
07:57 shorten dalek's url is at http://xrl.us/begeqt
07:57 dalek rakudo: c4f0f93 | (Moritz Lenz)++ | src/ (4 files):
07:57 dalek rakudo: Merge branch 'master' of git@github.com:rakudo/rakudo
07:57 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​4f0f93b97c77aeadf36276a677033aead07dce5
07:57 shorten dalek's url is at http://xrl.us/begeqv
08:08 iblechbot joined #parrot
08:10 integral joined #parrot
08:13 dalek parrot: r36789 | allison++ | trunk:
08:13 dalek parrot: [install] Install pmc2c.pl in devel package.
08:13 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36789/
08:15 dalek parrot: r36790 | allison++ | trunk/MANIFEST.generated:
08:15 dalek parrot: [install] Install generated file lib/Parrot/PMC.pm in devel package.
08:15 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36790/
08:23 dalek parrot: r36791 | allison++ | trunk/tools/build/pmc2c.pl:
08:23 dalek parrot: [install] Also search the installed lib path in pmc2c.pl.
08:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36791/
08:27 alvar joined #parrot
08:29 bacek joined #parrot
08:38 Andy joined #parrot
08:49 dalek parrot: r36792 | allison++ | trunk:
08:49 dalek parrot: [install] Change generated perl constants to a loadable module of 'use
08:49 dalek parrot: constant' statements, instead of a fragmented hash declaration. This
08:49 dalek parrot: allows PCCMETHOD.pm to be used from an installed Parrot. Also resolves
08:49 dalek parrot: RT #42286.
08:49 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36792/
08:56 dalek parrot: r36793 | allison++ | trunk/MANIFEST.generated:
08:56 dalek parrot: [install] Install the generated PCCMETHOD_BITS.pm in the devel package.
08:56 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36793/
09:08 mikehh joined #parrot
09:42 masak joined #parrot
09:43 mikehh joined #parrot
09:58 kj joined #parrot
10:07 dalek parrot: r36794 | fperrad++ | trunk/tools/install:
10:07 dalek parrot: [install] minor cleanup
10:07 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36794/
10:21 cognominal joined #parrot
10:47 masak ok, when I try and build Rakudo on feather, Configure.pl incorrectly sets the BUILD_DIR to /home/duff/parrot. why does it do that?
11:19 chid moritz, what should my question be :P
11:20 moritz chid: whatever you want to know :) - maybe just repeat what you wrote in #perl6
11:21 chid fix on https://trac.parrot.org/parrot/ticket/167 anyone? win32 bug
11:21 chid 'this -> load_bytecode 'PCT.pbc' isn't working correctly
11:21 chid it can't find PCT.pbc
11:23 chid nope
11:23 dalek parrot: r36795 | fperrad++ | trunk/tools/install/smoke_languages.pl:
11:23 dalek parrot: [install] check more languages
11:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36795/
11:23 chid if it was fixed it wouldn't be a bug :p
11:24 chid online
11:24 chid All that was nothing compared to the frustration that was caused by a merge in Parrot that broke Rakudo, which has just moved to git, but there was no one on IRC who could tell me where is the latest Rakudo. So for half a day I was staring at a broken Parrot and a broken Rakudo.
11:24 chid there
11:24 chid just like me
11:33 masak chid: on behalf of those who caused it, sorry for the inconvenience. I would suggest that you wait around until the US wakes up.
11:42 chid nah, I don't mind
11:45 Gerd joined #parrot
12:14 kj rurban: ping
12:15 kj message rurban I got undefined reference errors when building Parrot on a freshly installed cygwin
12:15 purl Message for rurban stored.
12:16 AndyA joined #parrot
12:19 rurban_ joined #parrot
12:21 kj rurban: hi
12:31 AndyA joined #parrot
12:43 Ademan joined #parrot
12:56 UltraDM joined #parrot
13:03 UltraDM joined #parrot
13:06 Gerd left #parrot
13:57 Nom joined #parrot
13:58 Andrew_ joined #parrot
13:59 Nom Someone here can probably answer this in one word.. if writing a shared module (ala DBDI) for all parrot language implementations to use, does that need to be in the parrot format i've seen around (pmc i think)?
14:00 Coke It doesn't have to be a PMC, no.
14:01 Coke I don't think we have a good guide on "so you want to use parrot as glue" yet.
14:01 Nom Well, i'm somewhat familiar with perl6, having played with it, but I got the feeling that you couldn't use it to write a module which interfaced with C libraries...
14:02 Nom It looked like it was Perl => Parrot => C Libs ... can the Parrot bit be written in Perl ?
14:02 Coke one of the long term goals is that you could share libs like that, yes.
14:03 Coke (by writing them in, say, perl6. or tcl. or python)
14:03 Nom but sounds like no, not right now :)
14:03 Andrew_ Nom: is possible mostly, just not documented and not all the details are ironed out
14:04 gryphon joined #parrot
14:04 Coke If you're wrapping an existing C lib, NCI is the way to go.
14:05 Coke there are already examples of that. That gets you the ability to invoke the C funcs from inside parrot.
14:05 Coke there are some examples of this in the tree, but I don't know that there's a pedagogical one.
14:06 Nom Yup.. i'm staring at DBDI::Driver::SQLite3 as an example now
14:08 Whiteknight Once you have the NCI wrapper for the library, any HLL should be able to tap into that
14:09 Whiteknight hopefully the various compiler designers haven't made subroutine invocation so convoluted that it doesn't Just Work
14:09 Nom Ah, I see...
14:11 Nom Ok, I get it now... so the PMC wrapper only needs to be there for the basics... it's likely that if, for instance, I wrote an OO library which wrapped a C lib, that eventually that could be used by any implemented language ?
14:11 Nom ie. PHP code could "use" the Perl library by having it built in as a parrot namespace
14:11 kj it needs to be a PMC if you want to use it in Parrot assembly/PIR
14:11 Whiteknight unfortunately, the weight lies on the shoulders of the compiler designers: They individually need to make their compilers flexible enough to handle Parrot libraries
14:11 Coke or we'll have a way for PHP to say "give me this perl6 namespace"
14:12 Coke "spec out HLL interactions" is, IIRC, still on the TODO list.
14:24 rg joined #parrot
14:37 gryphon joined #parrot
14:50 PacoLinux joined #parrot
14:58 gryphon joined #parrot
15:13 riffraff joined #parrot
15:14 NotFound Whiteknight: a possibility is to write some glue code that puts in the HLL namespace some subs.
15:15 Whiteknight NotFound: yeah, thre are a few workarounds. Eventually I would like to see a dedicated interface that can "autobox" low-level PMCs into HLL-mapped types, and register Parrot-subroutine names into whatever subroutine registration methods that HLLs use
15:17 NotFound Whiteknight: and also a way to make HLL PMCs available to other HLL more or less automatically, for modules intended for generic usage.
15:18 Whiteknight right
15:18 NotFound Here is the true parrot power :)
15:20 NotFound Did we already have a DBDI design documented?
15:20 PerlJam NotFound: I thought it was "like JDBC"
15:21 Andy joined #parrot
15:21 NotFound PerlJam: a bit vague
15:21 iblechbot joined #parrot
15:22 NotFound For example: does that means that all strings must be utf16 be?
15:22 NotFound Er, LE
15:23 PerlJam I'm not Tim Bunce, nor do I play him on IRC  ;-)
15:25 kj I have (still) failures when building parrot in a folder that has embedded spaces in its name. Anybody knows whether this can be solved?
15:25 kj It used to work, but not since 1 week or so
15:27 NotFound kj: i think that make that work in all current developing environments will be very hard.
15:28 kj NotFound: but it used to work..
15:28 kj it has for a long time.
15:29 NotFound kj: but now the Makefile must work in a variety of win32 environments, that makes the thing harder.
15:30 particle NotFound: roughly, the design for dbi 2 is to use the jdbc api
15:31 NotFound particle: but java has just one type of strings, that part at least need some specification.
15:46 kj anybody else has undefined references to glut functions (on linux)? I have them on cygwin
15:54 particle coke: ping
15:54 particle kj: i have cygwin, but don't have gl installed, i don't think
15:55 Coke particle: pong
15:56 Coke does anyone know who Igor is? Is he the grant recipient for the tcl/tk integration?
15:56 kj particle: how did you do releases, on windows or linux platform?
15:56 particle re: igor aka quelvar aka raf, i deleted quelvar user after his spam tickets
15:56 Coke particle: ok. that didn't work, so I guess I'll actually /talk/ to him.
15:56 particle coke: there are hundreds if not thousands of users defined at parrot.org
15:57 particle there is a bot somewhere signing up bogus accounts
15:57 japhb kj: your reference to glut pinged my IRC client ... I'm reading back, but is there anything you need help with?
15:57 kj japhb: I'm getting undefined references errors
15:57 Coke particle: we should relay that to whoever wanted us to switch to trac so they can work on it.
15:58 kj nopaste?
15:58 clunker3_ http://pasta.test-smoke.org/ or http://paste.husk.org/ or http://nopaste.snit.ch:8001/ or http://rafb.net/paste or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/
15:58 purl i heard nopaste was at http://nopaste.snit.ch/ (ask TonyC for new channels) or http://rafb.net/paste or http://poundperl.pastebin.com/ 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)
15:58 Coke DIE BOTS!
15:58 rg embed.h does not have a prototype for Parrot_compile_string. this is breaking the embed.t for me on freebsd/amd64 since the resulting pointer gets truncated to an int
15:58 particle :)
15:58 nopaste "kjs" at 193.1.104.7 pasted "glut errors" (19 lines) at http://nopaste.snit.ch/15634
15:58 Coke no wonder people don't come to this channel, it's 95% bots+perl6. =-)
15:58 kj japhb: I pasted the erros i got in cygwin
15:59 particle kj: i've done releases from both platforms. i have a unix-like environment for windows, so i have most of the command-line utilities that you need in linux
15:59 Coke particle: I see more like 3 dozen accounts on trac.
15:59 particle kj: that said, compile is much faster in linux, since i can use -j
16:00 kj particle: unix-like environment is cygwin?
16:00 Coke do you mean "www.parrot.org" ?
16:00 japhb kj: that's odd.  Looks like your GLUT variant was misdetected.  Can you paste the output of configure?
16:00 kj japhb: will do
16:00 particle kj: cygwin-- no!
16:00 Coke ah, yes, you do.
16:00 Coke so that was a non-sequitor on the raf thing.
16:01 particle kj: actually, git provides most of the commands (including tar)
16:01 particle coke: http://www.parrot.org/admin/user/user
16:01 particle yes
16:01 kid51 joined #parrot
16:01 nopaste "kjs" at 193.1.104.7 pasted "kjs' config" (406 lines) at http://nopaste.snit.ch/15635
16:02 kj particle: I was planning to do the release on cygwin.. on my laptop it works fine; on my (faster) office pc it doesn't
16:02 Coke particle: we are requiring e-mail verification when a visitor creates an account on www.parrot.org
16:03 kj afk # meeting
16:04 rg notfound, ping?
16:04 Coke particle: I wouldn't destroy anyone's accounts on trac again without talking to them.
16:05 particle yeah, i know....
16:05 japhb kj: OK, so it claims you have freeglut 4; freeglut is the most powerful GLUT, so it should handle bloody everything.  However, this looks a little odd to me:     set P0["opengl_lib"], "/usr/bin/glut32.dll -lglu32 -lopengl32"   Why would that not be '-lglut32'?
16:05 japhb Hmmm, I wonder if someone has been messing with the OpenGL detection again ...
16:07 * japhb rebases to examine further
16:07 kj japhb: seems like a typo, right?
16:07 japhb kj: more like a braino, but yes.  ;-)
16:07 NotFound rg: pong
16:08 rg notfound: interpreter.h does not export the prototype for Parrot_compile_string when embedding. this is breaking the embed.t for me on freebsd/amd64 since the resulting pointer gets truncated to an int
16:09 NotFound rg: C or C++ build?
16:09 rg c i guess
16:09 rg (whatever is the default ;))
16:09 NotFound For me it works for C and breaks with C++
16:09 rg are you on a 64bit platform?
16:09 NotFound I need to rework some headers, but It will take me some time
16:10 NotFound rg: 32, but I builded this morning on ubuntu amd64 and reported all test pass
16:11 rg also you need to be "unlucky enough" to have the pointer put at a memory location where truncating to int actually makes a difference
16:12 NotFound rg: but including interpreter.h is a temporary workaround, cmpile_string must be moved to embed.h or extend.h
16:12 kj japhb: I manually edited the makefile; -lglut32.dll doesn't work
16:12 kj same error
16:13 kj eh, -lglut32 i mean
16:13 rg including interpreter.h like you did is not enough. you'd probably also need to define PARROT_IN_CORE
16:13 japhb kj: hmmm, OK
16:13 NotFound rg: no, that exactly what must be avoided.
16:13 kj gotta go for 15 or so # meeting really starts
16:13 japhb kj, nod, I'll keep looking
16:14 rg sure. all that's missing is the prototype for Parrot_compile_string
16:14 rg maybe you just need to put that in embed.h or whatever suits you.
16:14 NotFound rg: I'll work on that as fast as I can. In the meantime you can TODO the test if it blocks you.
16:15 rg nah, doesn't bother me much. i just wanted to let you know.
16:15 NotFound rg: Ok, thanks
16:29 japhb Grrrr.
16:29 * japhb is bloody tired of JIT not working
16:29 japhb I really, really wish I could spare time to fix it, dangit.
16:34 elmex joined #parrot
16:37 Tene joined #parrot
16:38 NotFound In draft/pdd10_embedding.pod says: Parrot_compile_string - Note: this declaration should move from interpreter.h to embed.h
16:38 NotFound Then I'll start by doing that,
16:39 AndyA joined #parrot
16:40 rg note that i have no idea what/if something else is required to make embedding useable. i'm only interested in haveing the test(s) pass ;)
16:42 pmichaud hello.
16:44 NotFound rg: I'm interested in making embedding usable without breaking our own rules for embedding
16:47 gaz joined #parrot
16:57 masak pmichaud: when I try and build Rakudo on feather, Configure.pl incorrectly sets the BUILD_DIR to /home/duff/parrot. why does it do that?
16:59 pmichaud masak: perhaps that's the version of parrot it's finding somehow?
17:00 pmichaud masak: I think I need to re-do the order of Configure.pl probes for parrot.
17:00 masak ok.
17:00 masak after it finds that dir, it fails because it doesn't have read access to it.
17:00 masak pmichaud: do you have feather access?
17:00 pmichaud yes.
17:01 masak good.
17:01 pmichaud I'll try a build -- just a sec
17:01 pmichaud oh, yes, that's the issue.
17:01 pmichaud [pmichaud@feather ~]$ parrot_config build_dir
17:01 pmichaud /home/duff/parrot
17:03 pmichaud where is the parrot you want to be using?
17:03 pmichaud ../../parrot ?
17:03 purl ../../parrot is coded into languages/perl6/t/harness
17:04 pmichaud currently Rakudo's configure.pl looks for parrot/parrot_config, then an installed parrot_config, then ../../parrot_config
17:05 pmichaud I can potentially swap the order of the last two.
17:10 masak pmichaud: yes ../../parrot
17:10 purl ../../parrot is, like, coded into languages/perl6/t/harness
17:10 masak purl: no ../../parrot is <reply>
17:10 purl OK, masak.
17:10 pmichaud I'll try that.
17:11 Theory joined #parrot
17:13 pmichaud updated: c4f0f93
17:13 dalek rakudo: cf98887 | pmichaud++ | Configure.pl:
17:13 pmichaud oh, cf98887
17:13 dalek rakudo: Check for ../../parrot_config in preference to installed parrot.
17:13 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​f9888762ef722da84ff8b61c8791285bc9714b1
17:13 shorten dalek's url is at http://xrl.us/begftx
17:13 pmichaud see if that works any better.
17:13 * masak checks
17:15 masak it does indeed. pmichaud++
17:15 jonathan pmichaud: I have a Rakudo day this week. If you want me to spend some of it hacking on the setting stuff with you, let's arrange a day (tomorrow won't work for me, but Wed-Fri are fine)
17:17 pmichaud Fri won't work for me.
17:17 pmichaud Wed/Thu may be okay.  We had an unpleasant surprise here this past weekend, so I'm not sure what my schedule will be.
17:18 pmichaud let's plan for Wed, then.
17:18 pmichaud I often have other meetings on Thu mornings.
17:19 khatar joined #parrot
17:26 dalek rakudo: 924f2ba | pmichaud++ | perl6.pir:
17:26 dalek rakudo: Silently exit if passed '-le' argument from Test::Harness.
17:26 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/9​24f2ba4d036215f8e23a56eef3a5affb75c02d7
17:26 shorten dalek's url is at http://xrl.us/begfva
17:27 jonathan pmichaud: OK, Wed it is.
17:28 jonathan And sorry to hear about unpleasant surprise. :-(
17:29 pmichaud so far it looks like things are turning out okay
17:30 jonathan Ok.
17:31 jonathan Hope they continue to do so.
17:37 MagnusShortwave joined #parrot
17:48 dalek rakudo: 3da1733 | pmichaud++ | build/Makefile.in:
17:48 dalek rakudo: Remove commented-out "make Makefile" target
17:48 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/3​da1733db31198ba6a4d986f75b1ca3aa1950001
17:48 shorten dalek's url is at http://xrl.us/beggcf
17:48 dalek rakudo: 1f9086b | pmichaud++ | config/makefiles/ (2 files):
17:48 dalek rakudo: Remove obsolete config/ directory.
17:48 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​f9086bae5a1390dd0550d11c8146bcd8b9640e1
17:48 shorten dalek's url is at http://xrl.us/beggch
17:49 dalek rakudo: a6f4926 | pmichaud++ | Configure.pl:
17:49 dalek rakudo: Remove commented-out and obsolete calls to Parrot's 'reconfigure.pl' step.
17:49 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​6f4926f9415ecfd5c268df74d4b43b809a5e331
17:49 shorten dalek's url is at http://xrl.us/begghp
17:58 dalek parrot: r36796 | fperrad++ | trunk:
17:58 dalek parrot: [install] on Win32, don't use version in install path
17:58 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36796/
17:58 mikehh joined #parrot
18:04 Coke we are getting to the point where I'd really like to see every commit refer to a ticket. :|
18:04 Coke or a wiki page. or something to explain what thought was behind a given commit.
18:05 rg you mean better commit messages?
18:05 Coke no, though that would help
18:06 Coke I guess with trac it might even be close enough.
18:06 Coke for example, r36796 ... "but why not?"
18:07 Coke was that fixing a bug? Avoiding a problem? Just unnecessary but otherwise harmful?
18:07 Coke "unharmful"
18:08 Whiteknight Coke, that's a bad idea because it would require every fix to wait for a ticket to be filed about the problem first
18:08 szabgab I get Makefile:593: *** Recursive variable `MAKE' references itself (eventually).  Stop.  when trying to build parrot
18:09 Whiteknight better commit messages are always good though, they just shouldn't be forced to reference a ticket or wikipage
18:09 szabgab anything know or my system is broken ?
18:09 szabgab r36796
18:09 dalek parrot: r36797 | fperrad++ | trunk:
18:09 dalek parrot: [codingstd] fix SVN properties
18:09 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36797/
18:10 cotto Coke++ #dealing well with Igor
18:10 particle joined #parrot
18:10 szabgab now trying on another machine, where Parrot worked already
18:12 rg giving a reason for a change sounds like a good idea, but imho in the commit message would be enough.
18:13 nopaste "NotFound" at 213.96.228.50 pasted "Big patch: Forward declare parrot core types to avoid differences between core and extend/embed and move Parrot_compile_string to embed.h" (313 lines) at http://nopaste.snit.ch/15636
18:14 NotFound Someone can take a look at this thing?
18:14 szabgab hmm, I could compile on my regular machine
18:16 NotFound szabgab: check that make is available
18:17 szabgab yeah, without make that would have not run
18:17 szabgab but anyway, I updated Parrot, ran the whole thing again and now it worked
18:17 NotFound Uh, no, if not available the Makefile cannot be executing X-)
18:17 szabgab maybe bad timing.
18:18 gryphon joined #parrot
18:25 NotFound Coke: are you feeding trolls?
18:32 rg notfound: i can tell you that your patch fixes my test failure.
18:32 Coke NotFound: no.
18:33 clunker3 joined #parrot
18:33 Coke Whiteknight: knowing why people are changing code is not a problem.
18:33 Coke (shouldn't be forced to reference) Given all the crap we have to do for coding standards, I think this is pretty lightweight, actually.
18:34 Coke also: you can update tickets. you can't update commit messages.
18:34 Whiteknight we do put up with a lot of codingstd, but that doesn't mean we should just shovel more stuff on top of that
18:34 NotFound rg: good
18:35 Whiteknight if there's a change that needs to be made in the code, and there isn't a preexisting ticket for it, I'm going to just make the change to the code and not reference any ticket
18:36 Coke Whiteknight: what change is it? why is it needed?
18:36 Whiteknight it will be explained in the commit message
18:37 Coke And I'm saying that often times, what people are doing for commit messages is insufficient, and there's no way to annotate them to fix that up with more information.
18:37 Coke in fact, on the face of it, the commit message for the change I mentioned seems reasonable, doesn't it?
18:39 dalek parrot: r36798 | fperrad++ | trunk/lib/Parrot/BuildUtil.pm:
18:39 Coke I'm not suggesting opening a ticket to say "fixing this typo."
18:39 dalek parrot: [codingstd] remove hard tabs
18:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36798/
18:39 particle commits should say *why*, not just *what*
18:39 particle coke++
18:39 NotFound I need some opinons: can I commit the changes in that patch right now, or will be better to wait until the new release is done?
18:40 particle NotFound: talk to kj, he's the rlm
18:40 particle *release manager
18:40 Whiteknight Coke: bad commit messages are one issue, and they can be resolved. Adding a level of required hoop-jumping and bureaucracy to it is a Bad Idea
18:40 dalek parrot: r36799 | fperrad++ | trunk:
18:40 dalek parrot: [codingstd] remove trailing space
18:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36799/
18:40 NotFound kj: ping
18:40 Coke Whiteknight: how can you, ex-post-facto, resolve a bad commit message?
18:41 Whiteknight Coke: you can't resolve a bad commit message. You can ask people to make better ones going forward
18:41 Whiteknight There are lots of bad commit messages in the repo now that won't get fixed even if we add a heavy annotations requirement to them going forward
18:42 Whiteknight And with it, we still won't have any better guarantee, because commit messages will point to woefully incomplete ticket descriptions or empty wiki pages
18:42 Coke Whiteknight: strawman.
18:42 purl strawman is a bogus but common debating style.
18:42 Whiteknight if you want better commit messages, tell people what makes a better commit message and enforce it
18:42 Whiteknight and links to other places does not make a good commit message
18:42 rg actually the docs say you can fix a commit message (though i've never tried and it may require special privileges): http://svnbook.red-bean.com/nightl​y/en/svn.reposadmin.maint.html#svn.reposadmin.maint.setlog
18:43 shorten rg's url is at http://xrl.us/beggsk
18:43 Coke rg; if you muck with the repository itself, eys.
18:43 NotFound And also, you can mistype ticket numbers in commit messages. I do it a few times X-)
18:43 Coke we don't want to do that.
18:43 Coke Whiteknight: I think you're over-imagining the amount of work I'm suggesting here. I am probably under-imaginging it.
18:44 Whiteknight Any amount of additional work should be unacceptable.
18:44 particle we already have a style guide for commit messages
18:44 particle it's enforced culturally
18:44 Coke Whiteknight: like the amount of work we have to do to go back through and figure out what the hell people were thinking?
18:45 NotFound purl: mod_parrot?
18:45 purl somebody said mod_parrot was being worked on again.  ;) or at http://www.parrotcode.org/mod_parrot or at http://svn.perl.org/parrot-modules/mod_parrot
18:45 Coke You're right. it's unconcionable, and I would like to fix it. =-)
18:45 szabgab now rakudo is also broken for me
18:45 Whiteknight Coke: That's the problem of individual committers. A committer who writes lousy commit messages is also going to write lousy descriptions on wiki pages
18:45 szabgab error:imcc:syntax error, unexpected VAR, expecting '(' ('sub')
18:45 Coke Whiteknight: you're misunderstanding what I'm saying.
18:46 Coke I don't necessarily expect /the committer/ to write up more than what they would have in a commit message.
18:46 szabgab in file 'src/builtins/guts.pir' line 1111
18:46 Coke but (using yourself as an example), often comitters are working on larger chunks of work. Anything bigger than a breadbasket should already /have/ a ticket.
18:47 Coke so, if there's a "cleanup the GC" ticket, and you commit something in relation to that, it sure WBNI the commit message reflected that larger chunk of work.
18:48 Whiteknight isn't that the use of single-purpose branches, or commit tags?
18:48 Coke these are issues that are more relevant to pay-for projects where you have to keep track of effort spent to acheive a goal, but there are certainly PM lessons we could be using here.
18:48 particle sure would, and make tracking roadmap items easier
18:48 Whiteknight everything in the /cleanup_the_gc branch is going to be dealing with that topic
18:48 Coke Whiteknight: yes, a branch is one way of dealing with this problem.
18:48 particle whiteknight: you can just put 'TT #nnn' in your commit message
18:48 barney joined #parrot
18:49 Coke if you have a branch dedicated to a TT, you don't need to duplicate the documentation in each commit message.
18:49 particle correct
18:49 Coke I'd expect the mergeback to refer to the ticket, of course.
18:49 kj particle: (only TT #n's) I would prefer to have some kind of message indicating the commit, not only the TT number (because then I have to go browse to track on my poor connection that's a major PITA)
18:49 particle if you're committing to trunk, include the TT in the commit. if you're commenting on a TT, include the revision in the comment
18:49 Whiteknight you don't have to duplicate the documentation in each message anyway. That's a huge waste. You explain the commit
18:50 szabgab it must be something in the air, second time it compiled well
18:50 particle kj: of course, you should include the TT in the message, not replace the message with the TT
18:50 kj ok, right. Including it with the message.
18:50 kj ok then I misunderstood you there for a sec :-)
18:50 NotFound kj: please take a look at  http://nopaste.snit.ch/15636
18:50 dalek rakudo: 10909da | pmichaud++ | build/Makefile.in:
18:50 dalek rakudo: Make the fakecutable the default target.
18:50 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​0909da98af8a381935d8facbded8780fd52b5ca
18:51 shorten dalek's url is at http://xrl.us/beggtu
18:51 Coke Whiteknight: you still in school, or do you have a ``real'' programming job now?
18:51 particle -m "[t] todo/skip review; partially addresses TT #nnn"
18:51 Whiteknight Coke: "real" programmin job
18:51 kj NotFound: what's the patch about?
18:51 kj (looking at it now)
18:52 kj (iow, why would you like me to have a look at it? :-)
18:52 Coke Whiteknight: danke.
18:52 Whiteknight Actually, my job is a little light on actual "programming", so I get my needs satisfied through Parrot
18:52 NotFound kj: adds a header to forward declare core struct (pmc, string, interpreter) in order to use the same declarations in core and in extend/embed
18:53 NotFound kj: the question is: is a change to big to be done before the release?
18:53 NotFound s/to/too
18:53 kj NotFound: eh, I'm not sure what implications this has...
18:54 kj and, is this part of The Big Plan?
18:54 NotFound kj: well... maybe break mod_parrot, I don't know
18:54 kj well I'm already nervous with parrot stability as it is :-S
18:54 NotFound kj: ok, I'll wait and ask about it in #ps
18:54 Coke if there's a doubt, I'd recommend waiting until post releas.
18:54 Coke NotFound: or that. =-)
18:55 Coke reminds me, I have a TT from just before the /last/ release where I posted a patch that c. didn't apply.
18:55 kj major problem right now for me is the undefined references on my cygwin install; hope it's still ok on my laptop...
18:57 particle kj: i'm rebuilding on cygwin atm
18:57 particle currently in pmc2c
18:58 NotFound kj: Do you have a name for the release? If not, I suggest 'Negative NaN' X-)
18:58 dalek tracwiki: v13 | coke++ | NewParrotDeveloperGuide
18:58 dalek tracwiki: https://trac.parrot.org/parrot/wiki​/NewParrotDeveloperGuide?version=13
18:58 shorten dalek's url is at http://xrl.us/begguz
18:58 kj NotFound: I kinda had something already, but still pondering whether it's any good
18:58 kj (and have to check some things ;-)
19:01 geof joined #parrot
19:03 Coke Infinoid: does the tracwiki include the commit message?
19:05 jrockway joined #parrot
19:05 particle kj: my cygwin builds, testing now
19:06 * Coke tests feather.
19:06 kj particle: great.
19:06 particle i'm smoking cygwin and msvc atm
19:08 kj particle: do you think releasing from a cygwin platform would be advisable?
19:08 particle kj: i don't see why not
19:08 kj ok, just checking whether there's any particular reason not to
19:08 particle just provide the tarball to folks on other platforms for sanity check before releasing it
19:09 particle offer it as a temporary upload on ftp.parrot.org, or use some other file-sharing util
19:09 kj particle: ok, how do I access ftp.parrot.org?
19:10 kj attempting now...
19:10 particle iunn, i've never done it
19:10 Coke kj: one of the directors can email you the connection information. I presume we must share this with any RMs.
19:10 kj ok. I don't have access to that
19:10 kj not with my passwords anyway
19:11 particle yeah, you'll need the password
19:11 kj i figured, perhaps with trac pw
19:11 particle no, totally different account and pw
19:11 NotFound Created TT #343 with the patch
19:12 Coke www == drupal, trac == trac.
19:12 Coke and ftp is probably neither of those.
19:12 Coke I don't see an email with password information; I'd hit up parrot-directors@whateverlisthostis.org
19:14 kj Coke: doesn't that email arrive in YOUR box? :-P
19:14 kj being vp and all
19:14 Coke Yes, but it's the other 4 who might be able to help.
19:14 kj oh hehe
19:15 Coke (pretty much, it's allison, if we have the info collectively, it's through her.) I can open a ticket with OSU, and remote into the box, but that's pretty much the limit of my super VP power.
19:15 Theory joined #parrot
19:15 Coke hell, jerry's the president, bug him. he clearly outranks me.
19:15 particle i have the info
19:15 particle just found it
19:15 kj no contact info on the site, btw
19:16 Coke kj: I think there's a ticket about that.
19:16 kj particle: what's the protocol, is it as simple as emailing me and I promise not to tell anybody?
19:16 allison yes, we only have one login to the ftp server
19:16 Coke allison: BOO!
19:16 particle kj: yes, the protocol is ssh :)
19:16 allison no, it's not actually an ftp server, it's just a staging zone
19:17 * kj doesn't get what a staging zone is...
19:17 allison that is, we have one account to drop files in a virtual directory structure, and run a command that triggers the mirroring
19:17 allison OSU is running a whole host of mirrored ftp servers
19:17 * Coke is remided to open a ticket about docs.parrot.org
19:17 kj allison: but it /can/ be used to do temporarily file distribution?
19:18 allison kj: temporary file distribution?
19:18 kj allison: yes for tmp. uploading a release candidate
19:19 allison kj: I suppose so, as long as it's off in a special directory
19:19 dalek parrot: r36800 | whiteknight++ | trunk/docs/book/ch03_pir_basics.pod:
19:19 dalek parrot: [Book] Update section on Filehandle PMC, adding examples that show how the various methods relate to IO opcodes.
19:19 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36800/
19:19 allison kj: though you can just upload a temporary file to www.parrot.org
19:19 particle allison: kj wants to release from cygwin. to be sure the distro works on other platforms, i suggested he make the tarball available for testers before the actual release
19:19 kj how's that done for previous releases?
19:19 allison kj: which might make more sense
19:19 allison kj: we don't generally do release candidates
19:20 allison or, people just put up a file on their private web server, or on feather
19:20 kj allison: but I'm a bit of a noob regarding this, so I'd like to be sure not to mess up
19:20 allison kj: yes, it's good to get platform testing
19:20 kj speaking of which, what's the main purpose of PLATFORMS file?
19:21 particle to let users know if parrot is expected to work on their platforms
19:21 allison kj: to record the status of parrot compiling on various platforms
19:21 kj ah ok.
19:22 allison it's more for developers, to know what's expected to work
19:23 kj perhaps I'm overlooking, but can't seem to find anything on the www.parrot.org to upload a file
19:23 allison kj: you create a page, and attach a file
19:25 kj allison: found it, thx. Sorry for asking many questions
19:25 allison kj: I'd put it in http://www.parrot.org/scratch/parrot
19:25 allison (Scratch pages)
19:25 Coke if you're looking for a temporary place to stick something people can download, use feather.
19:26 kj Coke: you reckon that's a better option?
19:26 Coke if it's temporary and you have a feather account? very likely.
19:26 kj Coke: good question; if I do I've never used it
19:26 Coke feather?
19:26 purl somebody said feather was http://feather.perl6.nl/ or a perl6 community development server
19:26 Coke feather is also run by juerd.
19:26 purl okay, Coke.
19:26 Coke juerd?
19:26 purl well, juerd is root or at http://juerd.nl/ or mailto:juerd@juerd.nl
19:27 Coke so I have ~coke/public_html, e.g. on feather.perl6.nl, which maps to...
19:27 Coke http://feather.perl6.nl/~coke/
19:28 kj Coke: no feather account. Should I contact juerd?
19:28 kj .. to get one?
19:28 allison as long as you delete the file right after, either should be fine (I don't have a feather account either, but I do have a private webserver)
19:29 Coke kj: yes, juerd holds the keys.
19:30 allison kj: in the interests of timely release testing, best not to wait for juerd before distributing a test tarball
19:31 particle agreed
19:31 Coke I'd rather not post it on www.parrot.org without some advance thought about if/how we're going to do RCs in the future.
19:31 Coke Seems it would be much better to go the ftp route if we're going to do that.
19:31 allison kj: this time round, just create a Scratch page on www.parrot.org (quick, cheap)
19:32 Coke ok, nevermind me.
19:32 allison Coke: definitely don't post a main page on Parrot, but Scratch pages are just a developer sandbox
19:32 particle coke: we can delete the scratch pages and attachments
19:32 kj allison: I was planning to do a dry run tonight.
19:32 kj (once I'm home)
19:32 allison kj: sounds good
19:33 allison Coke: and agreed on thinking more about an RC process (if we're going to have one)
19:33 particle successful smokes with 'ccache cl' and 'ccache gcc' on windows/cygwin
19:33 Coke feather fails 'make test'
19:33 kj particle: yes on my msvc, smoke was successful
19:33 kj particle: thanks for testing
19:33 particle coke: fails? or bonus tests?
19:33 Coke both.
19:34 * particle has 4 bonus tests
19:34 Coke running make smoke now.
19:34 Coke yes, because the (&*#$ native_pbc tests are not properly TODO'd everywhere.
19:34 Coke Which is more a rant at the fact that we let it go this long rather than at poor rurban who is trying to make them work again.
19:35 allison Coke: yes, but the day before a release is definitely the time to re-"TODO" them
19:38 NotFound allison: can you take a quick look at TT #343 ?
19:39 allison NotFound: sure
19:40 Coke I'd rather have a release with passing todos than failures, sure.
19:40 Coke smoke report for feather here:
19:40 Coke http://smolder.plusthree.com/app/pu​blic_projects/report_details/18087
19:40 shorten Coke's url is at http://xrl.us/beggyt
19:41 Coke Could be related to a previously installed parrot.
19:42 kj I don't really get why these tests fail, as it's just a rename of functions
19:43 allison NotFound: looks good, thanks
19:44 NotFound allison: I'd like to commit it after the release, to have enough time to test in before 1.0
19:45 allison NotFound: yes, that's a good idea
19:45 NotFound allison: ok, thanks
19:46 allison Coke: yes, looks like an old copy of libparrot.so lying around
19:47 kj how to remove an installed parrot? seems not to be a make uninstall target
19:47 kj what's the normal procedure for that?
19:51 particle make sysadmin angry
19:53 NotFound kj: I think that the only way is lloking at what install does and deleting by hand
19:53 kj wouldn't it be an idea to do that manually?
19:53 kj eh
19:54 kj automatically?
19:54 purl rumour has it automatically is easier to set and forget, but manually is fine, too
19:54 kj or am I asking for too much? :-)
19:54 NotFound kj: yes, preferably by creating an uninstall target
19:54 kj exactly
19:54 particle more likely a generated uninstall script
19:55 kj well, isn't it just the reverse operation of install?
19:55 particle the makefile isn't installed
19:56 kj but what if you just go to the parrot folder, where the Makefile still is
19:56 particle it's likely you won't have your build dir around after you've installed
19:56 particle or, the makefile and install dirs may have changed
19:56 NotFound particle: that will be more adequate for package creators and maintainers
19:57 allison kj: yes, has to be done manually
19:57 allison uninstalling is a task for packaging tools
19:57 allison not for the core distribution
19:57 NotFound Uninstalling is a risky business, you must be extremely carefull when automatically deleting things as root
19:58 allison (that is, there are already good solutions out there, so don't reinvent)
20:01 cjfields joined #parrot
20:02 Coke allison: I submit that the test is broken if it picks up the installed parrot.
20:03 allison Coke: no, it's the installed parrot that's broken
20:03 Coke kj: there is an RT for creating an uninstall.
20:03 allison Coke: reject it
20:03 Coke I /opened/ it.
20:03 allison :)
20:03 allison okay, I'll read it first, RT #?
20:04 kj allison: the server listed in release_man_guide doesn't respond
20:04 chromatic joined #parrot
20:05 allison ah RT #43384
20:05 Coke looks like you rejected it on the sly.
20:05 allison kg: which server?
20:05 purl I'll get you, darkuncle's pretty! And your java app too!
20:05 kj ftp-osl.osuosl.org
20:06 Coke ah well. kj, if you want an uninstall target, check out:
20:06 Coke http://rt.perl.org/rt3/Tic​ket/Display.html?id=43384
20:06 Coke and then move on. =-)
20:06 allison ah, I guess I did
20:07 allison kj: are you trying to connect to it using an ftp client?
20:07 kj allison: yes
20:07 allison kj: because it's an ssh account, not an ftp account
20:07 kj tried scp, sftp
20:07 kj mm ok. somehow I thought a putty session *is* ssh...
20:15 allison kj: I was able to ssh just fine now
20:15 allison kj: try ssh under cygwin
20:15 allison kj: and don't for get to specify the user
20:16 allison (-l flag)
20:20 kj allison: just ssh'ing ; works fine, thanks
20:20 kj (just making sure everything works)
20:20 rurban_ joined #parrot
20:23 NotFound The second test in embed.t fails on freebsd/amd64, in c++ build, and maybe other cases, and there is no easy fix now. Must I skip or todo it before the release?
20:25 blair_sutton joined #parrot
20:29 allison NotFound: yes
20:30 NotFound allison: skip?
20:30 purl skip is 2.15?
20:30 kj I think skip
20:30 NotFound Less confusing, I think
20:31 Coke in general, use todo.
20:31 Coke (skips are harder to find and fix.)
20:32 allison NotFound: depends on the nature of the failure
20:32 allison NotFound: TODO runs the test, and reports success if it fails, skip never runs the test at all
20:32 alvar joined #parrot
20:32 rg won't todo result in unexpected success on working platforms?
20:32 allison NotFound: so, if it's a segfault, or something similarly fatal, you probably need skip
20:33 NotFound In that case, the thing required to do is TT #343
20:33 ask_ joined #parrot
20:33 allison rg: yes, todo reports if it unexpectedly works (so should only be todo'd where failing)
20:33 ask- joined #parrot
20:34 NotFound Is C code, the segfault problem is not so disturbing as in pir, but in this case I think is better to skip to avoid confusions.
20:35 rg i would think so too, especially since you have a fix that will remove the skip right after the release.
20:36 allison NotFound: seems sensible
20:36 NotFound Ok, I'll skip it
20:44 dalek parrot: r36801 | NotFound++ | trunk/t/src/embed.t:
20:44 dalek parrot: [test] skip one embed.t test not ready for all platforms, TT #343
20:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36801/
20:47 Coke also, it's much harder to conditionally todo things in pir.
20:53 blair_sutton hi, can anyone give me some pointers on how to create anon subs using HLL. i'm declaring my sub as a declaration block then assigning it with a bind op. i would like treat the newly defined sub as a Var,
20:54 kj blair_sutton: an anon sub is .sub main :anon
20:55 bsdz ah right. I imagine I can bind it to a PAST::Var using PAST::Op.
20:55 bsdz Can I use a past:op to call the bound past:var?
20:56 kj (i was talking PIR syntax)
20:56 kj yes
20:56 kj you can combine all PAST nodes, whether they make sense... pretty much yes
20:56 kj PAST::Op (:pasttype('bind') ) I think
20:57 kj please check pdd26
20:57 kj docs/pdds/...
20:57 bsdz thanks, any PCT-based languages that implement this functionality simply?
20:57 kj squaak :-)
20:57 kj did you read the tutorial?
20:58 kj www.parrotblog.org, see on the right 'tutorial'
20:58 bsdz yes. read lots. probably read the whole lot
20:58 bsdz started with the blog
20:58 bsdz original code is based on squak
20:58 kj ok
20:59 kj it's also in languages/squaak
20:59 kj it's slightly updated version from the blog
20:59 Coke another way to figure out what's what is to find the equivalent perl6 code, and see what it generates at the AST level.
20:59 bsdz yes. saw that. the svn version is more up to date than the blog. the blog code no longer works btw
20:59 Coke s/perl6/any language using PCT/
21:00 kj bsdz: yes that's the problem with blog code. and it's a pain to update it I might add ;-)
21:00 bsdz yes i did a little reverse engineering before.
21:01 bsdz thanks though. i will take another look at the docs and reverse engineer rakudo again.
21:01 pmichaud one can also get an anonymous sub in PCT by using a PAST::Val node
21:01 kj bsdz: please if you have any difficulties, please keep track of them, and post them to the mailing list or something
21:02 kj useful information for updating docs
21:02 pmichaud my $past := PAST::Val( :value($block) );   # where $block is a PAST::Block node
21:02 bsdz yes will do. will send to parrot.dev when i get a chance.
21:02 kj bsdz: great
21:03 bsdz cool, thanks pmichaud
21:03 pmichaud the $block still has to be a member of some other structure to get generated, though.
21:03 bsdz ok
21:03 pmichaud the PAST::Val is just a reference to the block
21:05 bsdz can i just use a PAST::Op(:pasttype('call')) on the new PAST::Val to execute it?
21:05 pmichaud yes.
21:05 bsdz great thanks
21:06 chromatic In function ‘memcpy’,
21:06 chromatic inlined from ‘cvt_num16_num8’ at src/packfile/pf_items.c:361,
21:06 chromatic inlined from ‘cvt_num16_num8_be’ at src/packfile/pf_items.c:513:
21:06 chromatic /usr/include/bits/string3.h:52: warning: call to __builtin___memcpy_chk will always overflow destination buffer
21:06 chromatic That looks scary.
21:06 Coke what are you running, linux? I'm sure OS X doesn't do that. =-)
21:06 * Coke hides.
21:07 chromatic Mac OS X will eventually provide a single header, called "oohshiny.h" with a single function called "NSPushCandyColoredButton()".
21:07 Coke I am drooling in anticipation of how shiny that button is!
21:07 chromatic Your words scratch the back of my eyes.
21:08 * Coke grins.
21:08 NotFound I'd like a function dowhatiwantnow(fast);
21:08 idemal joined #parrot
21:08 chromatic It will only work if you are The Steve.
21:08 NotFound Wonder
21:09 tetragon joined #parrot
21:09 * Coke ponders writing a (not for parrot) codingstd test that requires tabs instead of spaces.
21:09 * Coke feels a little evil about this.
21:10 NotFound long double d?
21:10 NotFound Code like that can be used unconditionally?
21:10 * kj goes home for food. back later on a crap connection
21:11 chromatic Hm, there are lots of hardcoded type sizes in that file.
21:11 NotFound We don't have a has_long_double or something?
21:11 chromatic What could possibly go wrong with that?
21:11 particle don't touch that code! it worked yesterday. mostly.
21:12 NotFound chromatic: that sizeof long double is less than 16, or that the compiler can't check the size because of the cast
21:12 dalek rakudo: 2ff60d5 | pmichaud++ | build/gen_parrot.pl:
21:12 dalek rakudo: First draft of script to automatically download and build
21:12 dalek rakudo: required version of Parrot.
21:12 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​ff60d506b11f88bd45bf143e4fe14f2eb1d4bf0
21:12 shorten dalek's url is at http://xrl.us/beghce
21:13 NotFound The compiler or whatever tool emits that diagnostic.
21:13 idemal joined #parrot
21:13 chromatic Hey, when my compiler says "This is always going to overflow", I'll take a compile time warning of crashy-crashy and fix it.
21:14 NotFound unsigned char b[8];
21:14 NotFound Looks like the compiler is right
21:15 idemal joined #parrot
21:16 chromatic Yeah, trying to pull 16 bits of data out of an 8 bit unsigned char is a recipe for disaster.
21:18 NotFound I don't understand what those functions are trying to do
21:19 chromatic I'm not sure they're even used.
21:20 dalek parrot: r36802 | chromatic++ | trunk/src/packfile/pf_items.c:
21:20 NotFound *dest = (double) d; .... but * dest is an unsigned char
21:20 dalek parrot: [src] Fixed compiler warnings about memory overflows during memcpy().  I'm not
21:20 dalek parrot: sure these are correct, but they don't warn and all tests pass.  If they're
21:20 dalek parrot: still wrong, that's because the code assumes too much.
21:20 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36802/
21:20 Coke (unused) fixin's too good fer 'em.
21:21 NotFound There is a comment: TODO: test , so maybe is not tested at all
21:25 * rg is still stumped by coke's supposedly easy ;) jit problem.
21:25 rg anyone with jit knowlege around?
21:26 NotFound rg: Basic, advanced or guru level?
21:27 rg not sure. the sinh math op fails in jit (most of the time). atan works, so do both with slow core.
21:28 rg i really don't see any difference between calling Parrot_sinh_n_n or Parrot_atan_n_n
21:29 rg also, how would running in jit cause the sinh function to behave differently to running with slow core?
21:29 NotFound Someone fixed atan?
21:30 rg this is not a border case and also on freebsd, so it should be fine.
21:30 dalek rakudo: 12b2b39 | pmichaud++ | build/Makefile.in:
21:30 dalek rakudo: Add a draft "make parrot" target.
21:30 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​2b2b39e8ebcd8d7540f3423c4f725e93341892e
21:30 shorten dalek's url is at http://xrl.us/beghdj
21:30 NotFound rg: try dropping the inline qualifier in the opcode
21:35 rg no difference
21:36 rg i can't say i'm seeing a difference in the generated code, either :(
21:37 NotFound rg: What's the result that sinh gives?
21:38 NotFound ok 13 - atan2 # TODO broken under JIT TT #201
21:39 NotFound Strange
21:39 rg NaN
21:40 rg it's sinh that's broken, not atan. i've only been using atan to compare the calls. i'll use any other op (that is not replaced with a cpu instruction)
21:41 NotFound rg: but atan was broken in the same way
21:46 rg interesting. i wonder who fixed it and what they did
21:47 rg also TT #38 claims it's still broken.
21:47 NotFound rg: and was, I've seen the todo passed rigth now
21:48 GeJ Good morning everyone
21:49 cjfields joined #parrot
21:50 rg does the trans_10 test pass for you (that's sinh)
21:51 rg btw. my source still has the inline qualifier on both atan ops.
21:51 rg and it still passes.
21:51 Whiteknight joined #parrot
21:51 cjfields rakudobug?
21:51 purl rumour has it rakudobug is mailto:rakudobug@perl.org
21:52 rurban I see you were talking about me :)
21:52 rurban just came back from a business trip
21:53 NotFound rg: the inline was a thing I tested with the atan2 problem and worked, but was just a test, not commited
21:53 NotFound sinh works for me in linux i386
21:54 rg this is all very strange
21:54 rurban memcpy(&d, src, 16); should copy the whole long double to dest, not just a single byte
21:56 rurban But I see lot of fixing potential there, indeed. #ifdefs for the 16-byte + 12-byte long double versions to make that clear e.g.
21:57 rurban and *dest = (double)d is indeed wrong.
21:57 NotFound rurban: the assignment after the memcpy uses one byte
21:58 NotFound That one, yes
21:59 kjs_ joined #parrot
22:01 rurban http://smolder.plusthree.com/app/public_pr​ojects/report_details/18087#first_failure is definiteoly caused by a libparrot in /usr/local/lib. wrong linker line. I fixe that months ago but the fix was rejected
22:01 shorten rurban's url is at http://xrl.us/beghgg
22:02 rurban cc  -L/usr/local/lib -Wl,-E  t/src/extend_13.o src/parrot_config.o -o t/src/extend_13 -Wl,-rpath=/home/coke/sandbox/parrot/blib/lib -Lblib/lib -lparrot -lpthread -lm -L/usr/lib  -licuuc -licudata -lpthread -lm -lnsl -ldl -lm -lcrypt
22:03 rurban *dest = (double)d  also uses the stacklocation for the double, not the dest ptr. completely wrong
22:06 particle1 joined #parrot
22:07 NotFound Something completely wrong and all test pass => need better tests
22:07 rurban nope. this section is todo'd
22:08 rurban I know that it doesnt work yet. I just wanted to get the needed functions in, and fix it later.
22:09 rurban and a compiler which knows about double and long double (if 16 or 12) should be able to cast it properly. We don't need to do that by hand then. that was the idea
22:09 rurban but I need a bit more time for that
22:11 rurban but the only current problem I see is -L/usr/local/lib at the front. it needs to be moved to the back
22:11 cxreg joined #parrot
22:13 rurban r36802 is completely wrong we can let it in, I'll fix it later
22:15 rurban And I know now why the compiler complained. It only knew about 12-byte long double (intel x86), so writing 16 was too much. needs to be #ifdef'ed
22:16 rurban I can fix that tomorrow
22:19 Andy joined #parrot
22:21 japhb kj: I'm back for a bit.  Are you still having problems with GLUT on your cygwin?  (FWIW, I had no problems other than the usual NCI JIT FAIL on Linux/i386)
22:21 kj japhb: i'm home now working on my laptop
22:21 kj will rebuild now
22:21 japhb kj: thx
22:22 kj on my laptop: implying that this is a different wc than at work, where I had those gl troubles
22:24 japhb kj: fair enough -- but is the environment basically the same? (cygwin on same base version of Windows, with same arch of CPU)
22:25 chromatic I still don't understand how you can read 12 or 16 bytes from an unsigned char.
22:25 kj japhb: I think different cpu, but same setup: win xp pro +cygwin
22:25 chromatic Except that it's an uc*.
22:26 japhb kj: I just meant, are they both 32-bit CPUs or both 64-bit CPUs, rather than one 32-bit and the other 64-bit?
22:26 kj japhb: oh sorry. Yes both 32
22:26 geof joined #parrot
22:27 japhb excellent, that rules out a whole class of problems ....
22:27 kj yes so at work I
22:27 kj .. ve got an ATI graphics card
22:27 kj can you still have opengl stuff then?
22:27 kj i'm not so familiar with that graphics stuff
22:28 ask- left #parrot
22:28 dalek rakudo: a8647e3 | pmichaud++ | Configure.pl:
22:28 kj parrot built fine on this setup (svn up'ed_), now running fulltest
22:29 dalek rakudo: Add "--gen-parrot" option to automatically download/maintain Parrot.
22:29 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​8647e3fbde6d696e0f2081a41b023539f7706ba
22:29 shorten dalek's url is at http://xrl.us/beghi3
22:29 dalek rakudo: 8695e4b | pmichaud++ | build/PARROT_REVISION:
22:29 japhb kj: yes ... OpenGL works really well on nVidia, fairly well on ATI, and from OK to crappy on Intel built-ins (but that last problem is true of Direct3D as well -- they're just not great)
22:29 dalek rakudo: Bump PARROT_REVISION to r36802.
22:29 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​695e4b18cada5882338d12a6cb94805d542e49a
22:29 shorten dalek's url is at http://xrl.us/beghi5
22:30 japhb There are times I wish IRC did embedded newlines in a single message, rather than bots having to send line after line manually and making a mess ...
22:30 japhb .oO( "My hindsight has better vision than yours ..." )
22:40 bsdz what does "=> \past" mean in my PAST target when using PCT?
22:49 jonathan That it's a reference to a node that has already been dumped earlier in the output.
22:50 bsdz ah okay. that makes sense. does it also tell me which one it is or perhaps i can work that out easily?
22:51 * Coke
22:54 Theory joined #parrot
22:54 jonathan bsdz: Unfortunately, not. :-(
22:54 jonathan I have wished for that before now too.
22:55 bsdz oh well. i'll soldier on. thanks
22:59 Limbic_Region joined #parrot
23:00 * Coke ponders how to test a darwinport against HEAD instead of 0.9.0
23:01 Whiteknight kj: Any testing you need from me tonight?
23:01 Whiteknight I'm running some errands soon, but I can set up some tests to run if you need them
23:01 kj testing is always welcome. What platforms are you again 64 bits right?
23:02 Whiteknight right now Ubuntu x86-64
23:02 kj I only have access to 32 bits
23:02 kj Yes, if you could do a fulltest, that'd be great.
23:02 Whiteknight okay, I'll set that up tonight and send you the results
23:02 kj it's 11pm here, so will go to bed in about 1 hour; but if you could post results to list or me personally, or a nopaste that's fine
23:04 Whiteknight okay, I'll get it started now
23:04 kj great, thanks!
23:04 Whiteknight okay, have to go!
23:04 Tene joined #parrot
23:05 kj anyone: "web.root" in tools/util/release.json still has value "www.parrotcode.org" ; should that be changed into www.parrot.org?
23:06 Coke I think that's just used to point at release announcements.
23:06 Coke which no longer have a URL that you can define ahead of time.
23:06 kj well it will be on the front page of parrot.org..
23:06 Coke right, but it's just posted as a story.
23:07 Coke You could just post that first and refer to the URL if needed.
23:07 Coke i'd have to dig through and see where else web.root is used.
23:07 kj mmm
23:07 kj it was still parrotcode.org for the last release, apparently
23:08 kj nopaste?
23:08 clunker3 http://pasta.test-smoke.org/ or http://paste.husk.org/ or http://nopaste.snit.ch:8001/ or http://rafb.net/paste or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/
23:08 purl nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels) or http://rafb.net/paste or http://poundperl.pastebin.com/ 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)
23:08 nopaste "kjs" at 193.120.116.182 pasted "failure on cygwin, (on winxp32)" (29 lines) at http://nopaste.snit.ch/15639
23:09 kj bleh. failures
23:12 dalek parrot: r36803 | coke++ | trunk/ports/macports/Portfile:
23:12 dalek parrot: An updated (but still non-functional) suggested port for 0.9.0
23:12 dalek parrot: review: https://trac.parrot.org/parrot/changeset/36803/
23:13 kj aah. the line number doesn't match in that failing test.
23:18 * Coke ponders adding a "-j 2" to the make args for the mac portfile. Would that be rude to users of lower end macs?
23:19 kj weird. When I run prove t/op/debuginfo.t, all is fine, but when the test is run as part of make fulltest, it fails (wrong line numbers)
23:19 rg kj: are you sure you're running it with the right core?
23:20 kj rg: ... no. what do you mean?
23:20 kj just 'make fulltest'
23:21 rg fulltest is running the tests with all the different cores
23:21 kj oh right
23:21 kj that's probably it
23:21 kj let me check
23:22 kj eh. how do you do that using prove?
23:22 kj .. specifying a different core?
23:23 Coke TEST_PROG_ARGS=-j
23:23 chromatic TEST_PROG_ARGS="--runcore=xxx"
23:23 kj ok, I never do this. export that?
23:23 rg or you run perl t/harness -f t/op/debuginfo.t
23:23 rg that's what i've been told last time i asked ;)
23:23 Coke there you go.
23:24 Coke anyone familiar with the process of building parrot for install, I'd appreciate a glance at TT#344
23:24 rg also see TT #281
23:27 kjs_ joined #parrot
23:30 kjs_ hee, i didn't leave!
23:34 TiMBuS joined #parrot
23:51 gravity joined #parrot
23:51 kjs_ joined #parrot
23:59 kj I got this warning: WARNING: ../lib/Parrot/Distribution.pm:284 File not found: inter_run.c
23:59 kj ISTR in the svn logs that inter_run.c was removed
23:59 kj (and it's not on my wc anymore)

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

Parrot | source cross referenced