Camelia, the Perl 6 bug

IRC log for #parrot, 2009-07-27

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:05 mokurai joined #parrot
00:06 theory joined #parrot
00:21 TimToady joined #parrot
00:24 rhr joined #parrot
00:25 theory joined #parrot
00:34 theory joined #parrot
00:56 * Tene remembers meeting kid51.
00:59 petdance joined #parrot
01:22 TiMBuS joined #parrot
01:43 Casan joined #parrot
01:46 Zak joined #parrot
01:55 skv joined #parrot
01:56 skv joined #parrot
01:58 skv joined #parrot
01:59 skv joined #parrot
02:00 skv_ joined #parrot
02:02 skv_ joined #parrot
02:03 wayland76 joined #parrot
02:21 dalek TT #877 created by johnewart++: Parrot fails to compile on OS X 10.6 with GCC 4.2 out of the box
02:22 petdance joined #parrot
02:41 janus joined #parrot
02:42 skv joined #parrot
02:56 skv joined #parrot
03:20 donaldh joined #parrot
03:29 dukeleto joined #parrot
03:29 Zak joined #parrot
04:29 dukeleto joined #parrot
04:40 wayland76 joined #parrot
05:28 wayland76 joined #parrot
05:33 hiroyuki_y joined #parrot
05:39 cotto joined #parrot
05:42 cotto Yay power outage
05:44 chromatic I had one last Saturday for 9 hours.
06:22 uniejo joined #parrot
06:32 japhb joined #parrot
06:42 wayland76 The occasional power outage is good for you :)
06:43 cotto I got to meet my neighbors.
06:45 bacek_at_work cotto: with chainsaw?
06:48 * cotto is confused
06:48 * cotto only needed a rusty butter knife
06:48 wayland76 cotto: bacek_at_work is suggesting derangement
06:49 wayland76 I see derangement can still produce arrangement
06:55 iblechbot joined #parrot
07:21 donaldh joined #parrot
07:32 szabgab joined #parrot
07:33 he_ Hmm, I see the ftp server doesn't carry the 1.4.0 bits yet.  Oversight?
07:35 cotto It should.  I uploaded them as part of the release process.
07:35 cotto Where are you looking?
07:37 cotto ftp://ftp.parrot.org/pub/pa​rrot/releases/stable/1.4.0/
07:37 he_ hmm, ok, I was looking in the wrong spot (devel)
07:38 cotto I thought that might be the case.
07:38 cotto I'm not sure if the current layout makes the most sense.
07:41 cotto But I also don't care a great deal.  There are more interesting problems to work on.
07:51 cognominal joined #parrot
07:51 he_ Sure, no problem -- I just didn't think to check, and have not figured out if there's a pattern to when a parrot release is designated "stable".
07:53 cotto he_, docs/project/support_policy.pod under Release Schedule
07:55 he_ aha, thanks.
07:56 cotto np.  Sometimes the problem is knowing where to look.
07:59 masak joined #parrot
08:02 cotto sleep
08:17 cognominal joined #parrot
08:21 mokurai left #parrot
08:58 mj41 joined #parrot
09:00 Zak joined #parrot
09:21 wayland76 joined #parrot
09:29 bacek joined #parrot
09:32 flh joined #parrot
09:37 MoC joined #parrot
10:01 bacek o hai
10:25 liangke joined #parrot
10:26 liangke masak: Hello. Long time no see again.
11:21 donaldh joined #parrot
11:24 uniejo joined #parrot
11:32 uniejo joined #parrot
11:34 jimmy perl6
11:34 jimmy perl6?
11:34 purl it has been said that perl6 is see Perl 6
11:34 jimmy Perl 6?
11:34 purl i think Perl 6 is amazing.
11:52 * moritz thinks so too
11:52 moritz purl: Perl 6 is also see http://perl6-projects.org/
11:52 purl okay, moritz.
11:55 jimmy Perl 6?
11:55 purl hmmm... Perl 6 is amazing. or see http://perl6-projects.org/
11:58 MoC joined #parrot
11:59 Psyche^ joined #parrot
12:04 Psyche^_ joined #parrot
12:53 snearch joined #parrot
13:14 snearch left #parrot
13:33 particle joined #parrot
13:41 dalek parrot: r40291 | coke++ | trunk/config/init/hints/darwin.pm:
13:41 dalek parrot: Avoid invalid compiler option on OS X 10.6 (gcc 4.2)
13:41 dalek parrot: TT #877, Courtesy johnewart++
13:41 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40291/
13:42 dalek TT #877 closed by coke++: Parrot fails to compile on OS X 10.6 with GCC 4.2 out of the box
13:44 HG` joined #parrot
13:45 Coke joined #parrot
13:45 Coke ~~
13:53 Coke on feather (a linux box) I have ulimit -c set to unlimited, and have a harness where perl is invoking parrot, and the parrot is segfaulting, but not dumping core. Shouldn't it generate a core file in this case?
13:57 davidfetter joined #parrot
14:05 PacoLinux joined #parrot
14:19 rg1 do you have write permission on the location where the core is supposed to be?
14:23 szabgab joined #parrot
14:41 Coke yes.
14:42 Coke (I'm just running this in my homedir.)
14:42 Coke s/in/under/
14:42 Coke and if something segfaults when running parrot directly, I typically get the core dump
14:43 rg1 could the harness be setting a limit?
14:43 rg1 or could the code change its cwd?
14:52 Coke it's not setting a limit on purpose, no. cwd is unchanged.
14:53 Tene Coke: I have vague memories of unlimited not doing what it should for core files.  try setting it to a large number instead.
14:53 Coke http://code.google.com/p/partcl/sou​rce/browse/trunk/tools/spec_info.pl
14:54 rg1 i'm out of ideas then
14:57 Coke Tene: I am having trouble finding a number that does anything other than unlimited. (ulimit -c <some number> in .bashrc ; ssh in to localhost. type ulimit -c :: unlimited)
14:58 rg1 you can use ulimit just like any other command. you don't need to do a new login.
14:59 Tene um.  ulimit won't carry over ssh.
14:59 Coke rg1: not if you've already set the ulimit
14:59 rg1 that too. you have to do it on the machine in question
14:59 Coke I am doing on the machine in question.
15:00 Tene Coke: ulimit is per-session, like environment variables.
15:00 Coke Tene: that's why I'm setting it in my .bashrc and getting a new shell.
15:00 Coke ... by ssh'ing back into the localhost.
15:00 Tene ... oh, I can't read.
15:00 Coke ah, found one. 999999 works. 9999999 ==unlimited
15:01 Coke sadly, nothing in 'man bash' to tell me what units that's in. =-)
15:01 moritz sizes are all in kb by default, methinks
15:01 rg1 that seems a strange limitation in bash. works just fine in tcsh
15:01 Coke ah, there we go. 1024-byte. so that should work. (thanks moritz)
15:02 Coke rg1: you can ratchet it down, not up.
15:02 Coke that way an admin can set a hard limit that the user can't override. (or the user can set a limit they themselves cannot override)
15:02 rg1 yeah, bash does that for me too. i didn't expect that.
15:03 moritz Coke: from 'man bash-builtins': "Values are in 1024-byte increments, except for ..." - but I agree that it's damn hard to find
15:03 Coke moritz: yah, found it, thanks.
15:04 Coke BAH. and now that I have a numeric ulimit... I can't run partcl's test suite because I updated parrot past a removal of a feature I am using.
15:04 Coke BLAH.
15:09 * Coke ponders updating the parrot-version I'm trying to be a range instead of a lower bound.
15:09 Coke s/trying/tracking/
15:11 hercynium joined #parrot
15:21 donaldh joined #parrot
15:50 rhoelz joined #parrot
15:51 rhoelz quick question about ticket #596: the folks on #perl6 say that such an idea would "increase GC pressure", so is the Parrot team holding off on that ticket for that reason, or is it just that no one's gotten around to it yet?
15:54 cotto rhoelz, it's just that nobody's had the tuits for it.  It's definitely a goal.
15:55 rhoelz mkay
15:56 cotto There's also the issue that such a patch would conflict with pcc_rewiring.
15:57 rhoelz what's that?
15:57 purl i heard that was how it does it.
15:57 cotto pcc_rewiring?
15:57 purl pcc_rewiring is, like, allison's branch, neh
15:57 rhoelz cotto: correct
15:57 cotto It's a branch allison's working on to optimize and simplify Parrot's calling conventions.
15:57 rhoelz I'm interesting in trying to make contexts PMCs
15:57 rhoelz ah
15:57 cotto I was just seeing if purl had anything useful.
15:58 cotto She's been working on it for a long time, but it's a hard problem with a low bus number.
15:58 theory joined #parrot
15:58 Tene cotto: allison's standard story is that she's going to start working on merging it to trunk shortly after oscon.
15:58 cotto I thought the branch was still explodey.
15:59 HG` joined #parrot
15:59 rhoelz do I need special perms to create a branch?
15:59 cotto If you have a commit bit, you're good to go.
15:59 cotto using a svn >= 1.5 makes life much simpler as far as branching and merging goes.
16:00 rhoelz cotto: will a Pugs commit bit work on Parrot? (I assume yes?)
16:00 cotto Nope.
16:01 cotto alternately, someone around here maintains a git tree, but I don't recall who atm.  It might be Tene.
16:02 HG` joined #parrot
16:02 rg1 i thought it was infinoid, but that was read-only
16:04 cotto It'd be nice to have something official.  I've been thinking about learning git by using it with Parrot.
16:04 cotto I hate being newbified but it's so shiny.
16:04 PerlJam yeah, Tene had one.
16:04 PerlJam I had one
16:05 PerlJam a coupel of other people had a git repo of parrot too
16:05 cotto bacek_at_work probably has one
16:05 rhoelz cotto: git *is* pretty awesome
16:17 whiteknight joined #parrot
16:19 Casan Alias: I'm still green when it comes to committing code to the project, but I'll get the hang of it soon.
16:19 Casan ignore the last message, key up, history is not always good
16:22 * jimmy likes 'history is not always good'
16:27 Psyche^ joined #parrot
16:34 Andy joined #parrot
16:35 Coke (conflict with pcc_rewiring) please don't avoid doing work to avoid impacting this branch that isn't moving.
16:40 Tene PerlJam: I have a git-svn tree.  I don't publish it anywhere, but I could send you a tarball if you'd like.
16:46 jimmy good night, #parrot.
16:47 whiteknight hello #parrot
16:50 ilia joined #parrot
17:04 desertm4x joined #parrot
17:10 payload joined #parrot
17:12 bacek joined #parrot
17:21 flh is DYNSUPER expected to work? it generates broken C code for me (inter->vtables[enum_class_something]->... where enum_class_something is not defined, because it's a dynpmc)
17:21 davidfetter joined #parrot
17:25 whiteknight flh: I've never heard of DYNSUPER before
17:27 flh ok, pmc2c.pod knows it, but it doesn't seem to match what I'm trying to do
17:27 whiteknight what are you trying to do?
17:28 flh SUPER(a,b,c) but for dynpmc
17:29 darbelo joined #parrot
17:30 flh I have a "value" dynpmc and a "block" dynpmc, "block" extends "value", and I'd like "block"'s init vtable to call "value"'s init
17:32 Coke I presume you're also doing something other than just invoking the super's vtable?
17:32 flh of course, otherwise the parent's vtable would be automatically called
17:33 flh I think I have to write something like that: Parrot_value_get_vtable(INTERP)->init(INTERP, SELF); ("value" is the parent dynpmc)
17:35 Coke Have you tried just "SUPER" ?
17:36 HG` joined #parrot
17:36 flh this was my first try :) the generated code looks like that: interp->vtables[enum_class_value]->init(interp, pmc);
17:37 flh but "enum_class_value" is not defined for a dynpmc
17:38 flh am I right to think that this is because dynpmc's only get this number at runtime (while regular PMC get it at compile time)?
17:42 mokurai joined #parrot
17:43 einstein joined #parrot
17:47 einstein i have got question about the INTERP->vtables[]
17:49 einstein i got this condition: INTERP->vtables[SELF->vtab​le->base_type]->_namespace != SELF->_namespace, but i don't know whether this is a bug or there is some reason for it
17:51 einstein but i can't find any documentation on this INTERP->vtables[], (it could be I oversee something)
18:00 japhb joined #parrot
18:02 chromatic joined #parrot
18:04 iblechbot joined #parrot
18:08 Coke flh: sounds about right.
18:08 Coke flh: DYNSUPER should probably do a named lookup of the parent instead.
18:09 Coke whatever the fix, if DYNSUPER is failing, please open a ticket.
18:11 flh so you're saying that DYNSUPER is (should be...) the equivalent of SUPER for dynpmcs? pmc2c.pod is not clear to me
18:14 hercynium joined #parrot
18:14 Coke It seems to make that the PMC "grammar" should not care if you're a dynpmc or a core pmc.
18:14 Coke s/make/make sense/
18:15 Coke but whatever the right way to fix it is, it's clearly broken now, so we could use a ticket.
18:19 flh I'll try to see if I can write a patch for that, and open a ticket anyway
18:21 particle i think the DYN means "if there are any superclass methods written in pir for this pmc, don't ignore them"
18:21 particle but i can't quite remember, as SELF and DYNSELF changed meaning a long time ago
18:22 HG` joined #parrot
18:22 particle perhaps looking at the svn log for the pdd will help, iunno.
18:28 flh do STRINGs need to be marked as being alive during a GC mark?
18:30 einstein i found a bug running this pir program i get "Foo9" insteads of the expected "Class" | newclass $P1 = "Foo9" | class $P3,$P1 | $P3 = get_root_namespace ["parrot";"Class"]  |   $P4 = get_class $P3 |  print $P4
18:32 einstein when doing this asociated class to parrot;class namespace get replaced with the class "Foo9" class obj
18:34 Coke particle: I cannot think of a case where you'd /want/ to ignore them.
18:34 Coke einstein: is that first line really newclass $P1, "Foo9" ?
18:35 einstein yes
18:36 jonathan flh: Yes.
18:37 flh jonathan, thanks, also I've just seen in string.pmc that STRING* can actually be coerced to PObj*
18:40 Coke einstein: odd. removing the class $P3,$P1 changes the result of the print, which I wouldn't expect.
18:40 Coke (since $P3 seems to be overridden by get_root_namespace)
18:41 jonathan flh: Yeah, they have the same initial bit of structure (see pobjh.h IIRC, but I'm rusty)
18:41 * jonathan -> food
18:43 einstein coke: yes remove the class $P3,$P1 indead changes the result
18:45 einstein newclass $P1 = "Foo9" | class $P6,$P1 | $P3 = get_root_namespace ["parrot";"Class"]  |   $P4 = get_class $P3 |  print $P4 -> gives "Foo9"
18:45 einstein newclass $P1 = "Foo9" | $P3 = get_root_namespace ["parrot";"Class"]  |   $P4 = get_class $P3 |  print $P4 -> gives "Class"
18:46 Coke though I wonder why we have 'get_class' and 'class'.
18:47 Coke ah. class would probably better be written as 'class_of'
18:49 einstein yes that is true
18:51 Coke looks like a bug to me; I'd open a ticket, be nice to have an explanation even if it's proper.
18:55 einstein the interp->vtables[type_num = 102 in my case]->_namespace points to the parrot;class after the newclass call, which is causing the problem
18:58 einstein ok
19:03 jrtayloriv joined #parrot
19:08 mberends joined #parrot
19:08 PerlJam einstein: it's probably just that since $P3 is a class, it uses that class' assignment vtable entry and since there isn't one, it defaults to copying some values but keeping the structure the same.
19:10 PerlJam (or something)
19:18 Coke PerlJam: but $P3 is being assigned to twice there.
19:19 Coke the p3 returned by get_root_namespace should override whatever was in the P3 before that.
19:19 einstein there is a interp->vtables[index to type]->_namespace and a interp->vtables[index to type]->pmc_class->_namespace, what is the difference between these to fields
19:20 einstein in the newclass op the interp->vtables[index to type]->_namespace keeps pointing to the original namespace and  interp->vtables[index to type]->pmc_class->_namespace get assigned "the correct namespace"
19:20 donaldh joined #parrot
19:22 ruoso joined #parrot
19:22 einstein there must be some reason why a pmc class has a seperate field for the namespace from the vtable entry, is there somebody who can tell me?
19:26 pyrimidine joined #parrot
19:27 pyrimidine nopaste: nopaste?
19:27 purl nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels) 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) or http://gist.github.com/ or paste or gtfo
19:31 einstein ok now i seems i solved the bug, i added " new_vtable->_namespace        = new_namespace;" in init_class_from_hash in class.pmc
19:55 szabgab joined #parrot
20:00 pmichaud_ I'm pretty sure the "class" opcode is deprecated, unofficial, or otherwise wrong.
20:00 pmichaud_ The correct opcode is "typeof"
20:02 einstein if that is true then is must be removed from the test cases
20:02 rhoelz left #parrot
20:05 coke joined #parrot
20:06 Coke it appears to be non-deprecated. I'm not sure how it would be unofficial.
20:07 * Coke finally gets annoyed enough to filter out the "tests are available" messages...
20:09 einstein typeof and class call exactly the same function the only differnce is the :object_classes meta data
20:10 pmichaud_ Coke: it doesn't appear in any of the pdds.
20:11 pmichaud_ as einstein says, they're two opcodes with nearly identical semantics.  And istr that allison said at one point (a very long time ago) that the 'class' opcode should go away.
20:12 chromatic That sounds familiar.
20:12 Coke if there's no ticket (and no entry in deprecated.pod), then it's not deprecated, despite allison's suggestion.
20:13 Coke and I would bet that that is not the only opcode not to appear in a PDD.
20:13 pmichaud_ that's why I also added "unofficial or otherwise wrong" in my statement above.
20:13 Coke O
20:13 Coke if we shipped it, it's official.
20:13 Coke wrong I can certainly get behind.
20:14 pmichaud_ okay.
20:14 Coke einstein: I would open a ticket.
20:14 Coke that will give folks who are not on IRC a chance to chime in.
20:15 Coke trac?
20:15 purl trac is a web-based software project management and bug/issue tracking system emphasizing ease of use and low ceremony. It provides an interface to the Subversion revision control systems, integrated Wiki and convenient report facilities.  http://projects.edgewall.com/trac/ or Python, SQLite and ClearSilver or killing killtrac or a bug-tracking tool or at https://trac.parrot.org/parrot/ or slow or REALLY slow
20:15 einstein ok i will do tomorrow, i will have to go to sleep in a moment
20:16 Coke k. thanks!
20:16 einstein i also solved an other bug, must I make i ticket for that one
20:16 * Coke finds out about killtrac from purl.
20:17 Coke killtrac?
20:17 purl killtrac is a perl replacement for trac that integrates RT, subversion, and MojoMojo (and CPAN Testers, and AnnoCPAN, and ...)
20:18 einstein ok i will find out tomorrow, it's the first time for me that I actually do work on an open source project
20:18 Coke http://dev.catalystframework.org/wi​ki/meta/tracpagestoport/beerrewards - catalyst's bounty page, with things spec out in terms of beer.
20:19 chromatic Congrats, einstein.  Welcome to the community.
20:19 dalek cardinal: 7171222 | (Joeri Samson)++ | src/classes/File.pir:
20:19 Coke chromatic, that's rude, t... oh, right, that's his nick.
20:19 dalek cardinal: The opcodes bsr and ret were removed in Parrot
20:19 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
20:19 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7171222f4b008c585b4ba495c778dfdfa24a46a6
20:20 * Coke wonders if he can trick Whiteknight into patching tcl to avoid bsr/ret
20:26 jonathan Coke: Noooo...this is making me want to go ahck on Catalyst. ;-)
20:26 Coke jonathan: I'll give you beer to hack on partcl.
20:28 jonathan Ooh. ;-)
20:28 moritz jonathan: we should also introduce Rakudo beer bounties ;-)
20:28 Coke fix the bsr/jsr stuff is worth a beer to me.
20:28 Coke "fixing"
20:28 moritz jonathan: 2 from me for the one who implements a REPL that prints the return value, and remembers lexicals
20:30 jonathan Coke: "fixing" as in, rewriting to not need bsr/jsr? :-)
20:31 chromatic Does fixing the REPL for lexicals require anything more than lifting their declarations to an outer scope?
20:31 Coke jonathan: bsr/ret were recently removed from parrot; partcl didn't stop using them time.
20:31 Coke "in time"
20:31 jonathan Coke: *nod*
20:31 jonathan chromatic: Well, and persisting that outer scope somehow I guess...
20:31 Coke they're only being used in two files, limited usage. getting 'make test' working against parrot svn head is sufficient.
20:31 Coke (partcl's make test)
20:36 dalek cardinal: 2581153 | (Danius Michaelides)++ |  (2 files):
20:36 dalek cardinal: Improve implementation of at, with tests. Simplify values_at by using at where appropriate
20:36 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
20:36 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/258115372bcf0d7342fd34dac9de5870f7a5e7d6
20:39 jonathan Coke: Well, I grabbed myself a checkout. But my Parrot and Rakudo trees have a tangle of uncomitted stuff that I want to get sorted out first...
20:42 Coke jonathan: there's a ticket at partcl.googlecode.org - feel free to claim it. Happy to give you a commit bit, also.
20:43 * Coke better put some $$ on his paypal account...
20:43 jonathan Coke: Yes, commit bit will probably also leave me feeling like I should use it. ;-)
20:43 jonathan I promise not to add a makefile target called...yeah...the obvious. ;-)
20:49 hercynium joined #parrot
20:59 mikehh_ joined #parrot
21:05 mikehh__ joined #parrot
21:10 dalek cardinal: ffa87cb | (Danius Michaelides)++ |  (2 files):
21:10 dalek cardinal: Improved implementation of [] and slice(), including supporting range and 2 arg versions.
21:10 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
21:10 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/ffa87cb68250cc8cc89ee5c187c252bc628cbea2
21:12 moritz it looks like the cardinal bus number has just increased, no?
21:13 PerlJam a tad, yes
21:14 Whiteknight joined #parrot
21:17 dalek cardinal: 035f6b2 | (Joeri Samson)++ | t/array/uniq.t:
21:17 dalek cardinal: Add 2 failing tests for uniq!
21:17 treed bus number?
21:17 purl rumour has it bus number is at http://c2.com/cgi/wiki?BusNumber or known as truck number
21:17 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
21:17 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/035f6b2f2d0ced4d1e380bb5e423074e01ec4b36
21:17 dalek cardinal: 634d0ba | (Joeri Samson)++ | t/array/uniq.t:
21:17 dalek cardinal: Added todo for failing test (but did not add issue yet)
21:17 dalek cardinal: Signed-off-by: Ted Reed <ted.reed@gmail.com>
21:17 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/634d0ba5649dca4f7945f2ac2578a31502b03f76
21:18 treed Aha.
21:18 treed There are three of us working pretty commonly on it this day.
21:18 dalek parrot: r40292 | whiteknight++ | trunk/DEPRECATED.pod:
21:18 dalek parrot: [DEPRECATED] remove mention of the stack stuff that was deleted yesterday
21:18 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40292/
21:18 treed these days
21:18 PerlJam treed++
21:18 treed Myself, joeri, and Danius.
21:19 treed Tene helps out, too.
21:19 treed Mostly just with advice.
21:19 treed We've got an IRC channel, and I'm considering starting a mailing list.
21:19 moritz cool stuff.
21:19 treed Mostly because joeri and danius are in a very diffirent time zone from me.
21:19 * jonathan is very happy to see more language development on Parrot.
21:20 * PerlJam too
21:20 * treed is happy to be doing something meaningful while he's unemployed.
21:20 treed Spent the first week or two replaying Final Fantasy 7.
21:25 Whiteknight love that game
21:29 * Infinoid spent a large amount of time in that game trying to pair obscure materia together to see what it would do
21:30 * Infinoid <-- OCD
21:32 treed I'm at the point now where I'm trying to get the gold chocobo so I can get knights of the round.
21:32 treed Which makes me bored, so I haven't played in a while.
21:33 Coke jonathan: where would you like the bit sent?
21:33 Coke (for now, contributions are under the same rules as parrot itself)
21:33 dalek cardinal: eed56d9 | treed++ |  (2 files):
21:33 dalek cardinal: Provide fixes for the recently added Array.uniq! test cases.
21:33 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/eed56d9f98b05a23700c4e7cb39c734ab0f0b0b4
21:34 * treed tries to figure out what to do when a pull request on github won't apply cleanly.
21:37 jonathan Coke: jnthn@jnthn.net is fine
21:38 jonathan treed: In my experience, try to read the git docs, find them incomprehensible, swear lots, and decide to have a beer instead...
21:38 jonathan treed: But I'm probably a bad example. ;-)
21:39 treed Haha.
21:39 * treed is going to try adding the fork to his local git as a secondary and merge from it and massage shit locally.
21:39 moritz usually sombody here or in #perl6 can help though :)
21:40 moritz the git docs are great when you mostly know what to do already
21:42 Tene treed: either that, or push back on the requester, saying "rebase against my tree, and send me a patch I can commit"
21:43 treed "push back"?
21:43 Coke jonathan: done
21:43 treed I actually have no idea why it's failing to apply.
21:44 moritz wrong line endings?
21:44 treed moritz: His other ones worked fine.
21:45 dalek parrot: r40293 | whiteknight++ | trunk/src/ops (2 files):
21:45 dalek parrot: [ops] add two new experimental ops, set_label and get_label. These will eventually replace get_addr/set_addr as they are used with Subs, Continuations, ExceptionHandlers, and the like. See TT #218 for more details
21:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40293/
21:45 Tene treed: I just mean socially.. just respond saying "Your patch doesn't apply cleanly.  Can you rebase it against my tree?"
21:46 dalek cardinal: 2d7a06c | (Joeri Samson)++ | src/classes/File.pir:
21:46 dalek cardinal: The opcodes bsr and ret were removed in Parrot
21:46 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/2d7a06c4a7145bcf5e6b67e60f37f5bc82dd9c0f
21:46 dalek cardinal: 7a79754 | treed++ | :
21:46 dalek cardinal: Merge branch 'master' of git://github.com/joeri/cardinal
21:46 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7a797543fae7143c89d052afe866c601e55acc05
21:47 moritz treed: make: *** No rule to make target `src/classes/Any.pir', needed by `src/gen_builtins.pir'.  Stop.
21:47 moritz that's what the current cardinal says here
21:47 Tene moritz: use rake instead.
21:48 Tene moritz: the make infrastructure in cardinal has been deprecated by treed's rake infrastructure
21:48 Tene the make stuff just hasn't been removed yet.
21:49 treed moritz: what Tene said
21:49 treed I have to add a clean target to Rake.
21:49 treed Then I'll remove the makefile thing.
21:49 moritz and then 'rake test'?
21:49 bacek joined #parrot
21:49 treed test:all
21:49 treed Will run the entire test suite.
21:50 treed test:stats will do that, and include stats about which issues are causing how many failures
21:50 Coke Whiteknight: if they're experimental, shouldn't they be in experimental.ops?
21:51 Coke msg Whiteknight if they're experimental, shouldn't they be in experimental.ops?
21:51 purl Message for whiteknight stored.
21:51 treed moritz: It takes 4-6 minutes on my system.
21:52 treed Usually around 5, but sometimes much shorter or longer.
21:53 moritz it uses only one CPU it seems
21:54 treed Unsure if it's possible to get Rake to use more.
21:55 moritz it doesn't have a -j option
21:55 treed Aha.
21:55 treed You declare multitask instead of task
21:55 treed And it'll do the prereqs in parallel.
21:56 Tene Nice!
21:57 dalek parrot: r40294 | jonathan++ | trunk/runtime/parrot/library/P6object.pir:
21:57 dalek parrot: [p6object] The ACCEPTS defined in proto-object was a bit liberal over what it accepted as matching the Any type, such that both Any  Object and Object  Any, which could tie Rakudo's multi-dispatcher in all kinds of knots when it tried to sort the candidates. This patch resolves this while also going to some effort to keep us interoperating with calls from other languages. We'll see how it flies.
21:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40294/
21:57 jonathan treed: Will you accept a patch for a leaves target? ;-)
21:57 treed Haha.
21:57 treed Only if rakudo will accept a patch for a love target.
21:59 jonathan Well, maybe after I've had a few more beers. ;-)
21:59 treed moritz: As soon as I finish merging joeri's stuff, I'll push a change to the Rakefile to use multitask for the multiple-test targets.
21:59 moritz treed: cool
21:59 * treed is curious to see how much faster it'll be.
21:59 treed I only have two cores.
21:59 treed But even a 30% improvement will save me a lot of time.
22:00 moritz rakudo's parallel testing brings about a 45% improvement on two cores
22:02 mj41 joined #parrot
22:06 dalek cardinal: d0183aa | (Joeri Samson)++ | t/ (6 files):
22:06 dalek cardinal: Do not concatenate integers to strings in tests
22:06 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/d0183aac067b4128e523b79b377be3d08107b4ff
22:06 dalek cardinal: c0c2cc9 | (Joeri Samson)++ | t/array/uniq.t:
22:06 dalek cardinal: Added todo for failing test (but did not add issue yet)
22:06 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/c0c2cc9a403edcf0300d05c93c9a1e5fb6c68392
22:06 dalek cardinal: 0d96626 | treed++ | :
22:06 dalek cardinal: Merge commit 'joeri/library'
22:06 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/0d96626d70a96d2d2df4176fe699bf2614e1152c
22:07 treed Yeah, next time I'm having him rebase.
22:10 Whiteknight Coke: good call
22:10 Whiteknight (I forgot about experimental.ops)
22:13 dalek parrot: r40295 | whiteknight++ | trunk/src/ops (2 files):
22:13 dalek parrot: [ops] on suggestion from Coke++, move the new opcodes to experimental.ops
22:13 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40295/
22:14 treed Awesome.
22:14 treed There's a bug with Rake and namespaces.
22:14 treed http://rubyforge.org/pipermail/r​ake-devel/2009-July/000775.html
22:14 treed Only recently discovered.
22:14 treed And I managed to find it right away.
22:15 PerlJam treed: Are you guys pulling in the standard ruby test suite to test cardinal?
22:15 PerlJam (Is there such a thing? I just assume there would be)
22:16 Tene PerlJam: The what ruby what what?  Rubyists don't believe in testing.
22:16 Tene Unless you make a DSL.
22:16 Tene I guess.
22:16 Tene More seriously, there is no standard ruby test suite.
22:17 moritz last time I looked, ruby had a ridiculous small amount of tests, and some assertions in the interpreter
22:17 treed Pretty much, yeah.
22:17 Tene There are a couple of third-party attempts, last I looked.
22:17 treed PerlJam: We're just writing our own tests as we go along.
22:17 Tene None of them are easy for a new compiler to use.
22:17 treed I think things would explode if we tried a full spec test.
22:17 treed Maybe in a year.
22:17 Tene the biggest one I found used rspec: http://rspec.info/
22:18 Tene which 1) requires some weird stuff, iirc
22:18 Tene and 2) the faux-dsl-ish thing makes my eyes bleed
22:18 PerlJam Tene: have you seen cucumber?
22:18 Tene I havne't.
22:19 PerlJam http://www.rubyinside.com/cucumber-t​he-latest-in-ruby-testing-1342.html
22:19 PerlJam get something to wipe the blood from your eyes.
22:19 Tene ... I'm at work right now.  I don't want my students to see me cry.
22:20 Tene also, gdiaf
22:20 Tene I mean, um, I don't think I like it very much, personally.
22:20 Tene >.>
22:20 PerlJam There's a railscast for cucumber too I think.
22:21 dalek rakudo: 53059d1 | pmichaud++ |  (4 files):
22:21 dalek rakudo: Correct Code.ACCEPTS and Regex.ACCEPTS .  [1,2,3] ~~ { ... }  should now work properly.
22:21 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/5​3059d1c5435fce71029ecfad98b0b6f350d5d1e
22:24 kid51 joined #parrot
22:24 treed Oh, god.
22:26 treed My system got very, very slow for a few minutes there.
22:26 treed multitask may not be a good idea
22:27 treed Not sure if the tests are getting done much faster overall.
22:28 treed So far all I can tell is that they're out of order.
22:28 treed Nope, took just as long.
22:28 treed Awesome.
22:29 * treed needs to get started on cleaning the floors like he was supposed to today.
22:38 rg joined #parrot
22:46 dalek TT #878 created by jkeenan++: Recent commits break buildtools_tests
22:50 Tene treed: um, how many processes were launched at once?
22:50 Tene Should be (number of cores)+1 at most
22:50 Tene if it just did all of them... yeah, that's gonna suck.
22:51 Tene treed: you could also look into fixing precompilation and then precompiling all the tests. ;)
22:51 Tene you'd still have to rebuild them for any parser changes...
22:54 dalek rakudo: 82b59c4 | jnthn++ | :
22:54 dalek rakudo: Merge latest.
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​2b59c45c398669a3bd72e25e07ee3527080fda5
22:54 dalek rakudo: d043978 | jnthn++ | src/pmc/perl6multisub.pmc:
22:54 dalek rakudo: Fix a bug in the narrowness analyzer for multi-dispatch (not being narrower is a necesary but not sufficeint condition for being tied).
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​04397830afd46041708e6695c3101712f1b34aa
22:54 dalek rakudo: 1f76da5 | jnthn++ | src/classes/Any.pir:
22:54 dalek rakudo: does in Object should delegate to .^does, not .^isa.
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​f76da51ea544fd2a71fd293b84611021afdbb38
22:54 dalek rakudo: c372689 | jnthn++ | src/classes/Role.pir:
22:54 dalek rakudo: A role has always been accepted by Any up until now, mostly by accident. Now make it really so. Turns out sigil based dispatch doesn't really work out without this.
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​3726899c8957de5966137d089c2bbb6acc73dc5
22:54 dalek rakudo: 1878aef | jnthn++ | build/PARROT_REVISION:
22:54 dalek rakudo: Bump PARROT_REVISION for p6object ACCEPTS fixes.
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​878aefc53f8ba114c2696dcb17fa253ce495139
22:54 dalek rakudo: 68d05a5 | jnthn++ | src/setting/Object.pm:
22:54 dalek rakudo: WALK assumed that Truee.ACCEPTS($x) could accept a $x !~~ Any. The smartmatching spec says otherwise. Correct it.
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​8d05a57750524899d1a04e3ab4523433c5b6b37
22:54 dalek rakudo: 9ea46ee | jnthn++ | :
22:54 dalek rakudo: Merge branch 'master' of git@github.com:rakudo/rakudo
22:54 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/9​ea46ee0c8eba48e6f9ab8a809dfbbf035b8fad2
22:57 treed Tene: I expect that it did them all at once.
22:57 Tene yeah, that's a bad idea...
23:00 treed I wonder if I could fork Rake and fix that.
23:00 Tene http://www.ruby-forum.com/topic/86952
23:04 treed Well, I'd just make a threadpool but keep the same multitask thing.
23:04 treed Just limit the amount of threads that can be going at any given moment.
23:09 mrsaturn joined #parrot
23:10 mrsaturn hi all
23:10 patspam joined #parrot
23:15 mrsaturn left #parrot
23:15 dalek decnum-dynpmcs: r121 | darbelo++ | trunk/src/pmc/decnum (2 files):
23:15 dalek decnum-dynpmcs: [DecNumContext] Take advantage of some library-defined constants.
23:15 dalek decnum-dynpmcs: [DecNum] add a get_exponent() METHOD, to get at the real exponent in a number.
23:15 dalek decnum-dynpmcs: Useful for debugging, as the exponent might differ from the one printed by the
23:15 dalek decnum-dynpmcs: get_string() VTABLE.
23:15 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=121
23:20 donaldh joined #parrot
23:26 dalek rakudo: ea667e8 | pmichaud++ | docs/spectest-progress.csv:
23:26 dalek rakudo: spectest-progress.csv update: 423 files, 12043 passing, 0 failing
23:26 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​a667e8db3d36da27479a7bbd7518ed308485b9b
23:37 dalek decnum-dynpmcs: r122 | darbelo++ | trunk/src/pmc/decint.pmc:
23:37 dalek decnum-dynpmcs: [DecInt] Make the divide() VTABLE, forward to the floor_divide.
23:37 dalek decnum-dynpmcs: Make the set_[string|number|integer]_native VTABLEs round the number correctly.
23:37 dalek decnum-dynpmcs: Copy-paste from DecNum while I figure out WTF is wrong with SUPER().
23:37 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=122
23:40 dalek cardinal: ecc08a4 | treed++ | t/range/each.t:
23:40 dalek cardinal: Change a TODO comment into an actual todo()
23:40 purl dalek: that doesn't look right
23:40 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/ecc08a4934d108ada1eaf8b105f6cf5b9cdc1c70

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

Parrot | source cross referenced