Camelia, the Perl 6 bug

IRC log for #parrot, 2009-08-25

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:03 * duk3leto love me some vim
00:03 * duk3leto *loves* me some vim
00:04 * darbelo resists the urge to link to xkcd.
00:05 Austin Have you been trying this for over half an hour?
00:06 bacek_at_work Austin: trying what?
00:06 purl trying is just as important as succeeding sometimes.  :)
00:06 Austin bacek_at_work: http://www.xkcd.org
00:07 Tene purl: msg allison When do you get to Redmond?
00:07 purl Message for allison stored.
00:07 paco joined #parrot
00:08 bacek_at_work Austin: I've got 3 t-shirts from xkcd
00:08 * Tene considers vim mandatory for editing text.
00:08 Austin Yes. But do you have the flowchart prominently displayed next to your monitor?
00:09 Tene Whiteknight: like many other things worth using, vim is optimized for the experienced user, at the expense of the beginner.  I've taught *many* people to use it, though, so I very much feel your pain. :)
00:09 bacek_at_work Austin: no. I can't even say which keys I press to do something complex. It's just in my fingers.
00:11 Austin Tene: if you'd stop making silly mistakes, you wouldn't need vim. You could use the editor that doesn't waste your VALUABLE time!
00:11 Austin Ed, man! > man ed
00:11 Austin "ed is the standard text editor."
00:11 darbelo Austin++
00:12 Whiteknight Tene: I'm sure it would be fine if I took the time to learn it, but I don't want to take the time
00:12 Austin http://www.gnu.org/fun/jokes/ed.msg.html
00:12 Tene I've tried it.  used it exclusively for a few days to see what it was like.
00:12 Austin ed++
00:12 Tene it didn't live up to the hype.
00:12 Tene vim is much more productive than ed
00:13 Tene I do have a co-worker who has EDITOR set to ed, though.
00:13 Austin Of course he does.
00:13 Austin Ed is the standard.
00:14 darbelo Tene: compare "which vi" and "which ed"
00:14 Tene Sure.
00:15 darbelo Then think which one will allow you to unhose the configuration if /usr becomes unmountable.
00:15 dalek partcl: r624 | coke++ | trunk/runtime/builtin/set.pir:
00:15 dalek partcl: cleanup PIR
00:15 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=624
00:15 dalek partcl: r625 | coke++ | trunk/ (2 files):
00:15 dalek partcl: Don't try to autovivify namespaces on [set]
00:15 dalek partcl: adds a test, passes at least one more spectest.
00:15 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=625
00:15 dalek partcl: r626 | coke++ | trunk/runtime/builtin/info.pir:
00:15 dalek partcl: [info body] shouldn't die on the empty-named sub.
00:15 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=626
00:16 Tene darbelo: A tool being useful sometimes doesn't mean it should be exclusively used.  I often use X, too.
00:16 Tene I even use (*gasp*) GVIM
00:16 Austin Well, there's your problem, right there.
00:17 Tene But I don't have a problem?
00:17 Austin The first step toward solving a problem is to admit you have it. If you can't make that step, we can't help you.
00:17 Austin Sorry.
00:18 darbelo Tene: I'm not advocating exclusive use of ed. I'm just recomending knowledge of enough ed to fix whatever is preventing you from mounting /usr/local and starting up emacs :)
00:18 Tene I don't want to stop!
00:21 Austin Actually, fwiw, I use Notepad++, which is a scintilla-based editor.
00:21 Whiteknight I use Notepad++ most of the time on Windows
00:21 Austin I used to use scite, but np++ offers tab drag-n-drop.
00:21 Austin I like an editor that starts up right now.
00:22 Whiteknight I tried using it on Wine when I switched to Linux, but didn't make the gap gracefully
00:22 coke joined #parrot
00:22 Austin No. Better to go with Scite for that.
00:22 * Coke hopes the random reboots are over.
00:22 Austin Which random reboots?
00:22 Austin Did you finally port linux to parrot?
00:22 Coke of my home machine.
00:23 Coke the room it's in is being painted.
00:23 Austin okay....
00:23 Coke allison++
00:23 Austin ?
00:24 Coke non-sequitor
00:24 purl uhh, heh, he said ass, dude
00:25 Austin s/sequitor/sequitur/
00:26 * Coke makes a note to run all his sends to austin through the spell checker.
00:26 Austin Nah, but that's a common glitch.
00:31 Austin So why does paint cause random reboots?
00:31 kid51 joined #parrot
00:32 Coke Austin: the computer was incautiously moved.
00:32 Austin Ahh.
00:33 Austin That doesn't sound random at all.
00:34 Whiteknight man, I can't figure out how to do anything in eclipse
00:35 Austin LOL
00:35 mokurai joined #parrot
00:35 Austin I generally have the same problem, Andrew.
00:35 Austin (Which is one reason I don't use eclipse.)
00:35 Whiteknight this is why I don't end up using more fancy tools, because I can't ever just do the simple shit I want to d
00:35 Austin I think you need to WTFM, so all of us can use it.
00:36 kid51 Eclipse was recommended to my a couple of days ago by DrMath for editing a file with a ton of global variables -- and some locals with the same names as the globals.
00:36 Austin Written in?
00:36 kid51 ... but I haven't tried it yet.
00:37 Coke I use eclipse for cold fusion, but nothing else.
00:37 Austin kid51: Eclipse has some text editors that are syntax aware. If your file is in one of those languages, it can tell the difference, and can replace, say, all of the "global Foo" symbols with gFoo. It's quite nice that way.
00:38 Austin But you have to know how to get it all set up, first, and of course it has to be a supported language.
00:38 Austin I'm pretty sure there's no pir-mode for eclipse, for example.
00:39 kid51 The code in question is Perl 5 ... very ugly Perl 5 ... originally by a certain Mr. Wall.
00:39 Austin Is it ugly-ugly? Or is it "I didn't think this was possible in any programming language" ugly?
00:41 kid51 Judge for yourself:  perldoc -m ExtUtils::ParseXS
00:42 Austin # use strict;  # One of these days...
00:42 Austin 'nuf said.
00:42 kid51 That's my current project:  make it work under use strict.
00:43 * kid51 installs Perl 5.10.1 on Darwin
00:43 darbelo It has 'XS' on it's name, that's a sure sign I won't be opening that file anytiem soon.
00:43 Austin I'm with you, darbelo.
00:44 kid51 What am I doing on my summer vacation?   Trying to work through 'perlxstut' tutorial.
00:44 Austin Hmm.
00:45 darbelo kid51: Be careful. xs can severly harm what's left of your sanity.
00:45 Austin I have a whole lot less sympathy when you tell me that you put your reproductive organ into the sausage grinder on purpose.
00:45 Whiteknight eclipse is turning out to be barely usable
00:45 Whiteknight I cant do anything
00:45 kid51 The ParseXS.pm code is much uglier than any of the Perl 5 in Parrot that I have worked on.
00:55 Whiteknight okay, I closed it. Cant deal wth eclipse
00:57 Austin Thank god.
00:57 purl amen to that
00:57 Austin The last thing we need is another layer of slatheron in parrot.
00:58 pmichaud today's xkcd.org is *perfect*
00:58 pmichaud I've already forwarded it to my dad.  :-)
01:03 Coke if I use .const 'Sub', there's no runtime cost on that, right?
01:04 pmichaud Shouldn't be.
01:04 Coke (ok; moving it up to the beginning of the enclosing .sub because it reads better, just making sure I'm not screwing myself.)
01:10 dalek partcl: r627 | coke++ | wiki/SpecTestStatus.wiki:
01:10 dalek partcl: Don't bother running this test for a stub.
01:10 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=627
01:14 Coke I'm kind of wierded out that my implementation of [chan] (which was just .return('')) passed several spec tests. =-)
01:15 dalek decnum-dynpmcs: r168 | darbelo++ | wiki (2 files):
01:15 dalek decnum-dynpmcs: Straight paste of my first (incomplete) draft of the DecNum documentation. This
01:15 dalek decnum-dynpmcs: needs more work, but it's going to have to wait until tomorrow, before I can do
01:15 dalek decnum-dynpmcs: any work on it.
01:15 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=168
01:17 darbelo Hmm, edits on the wiki spam the channel too. I wasn't expecting that.
01:18 kid51 darbelo:  Feature, not bug, IMO.  It pokes people to actually go and read the wiki, which they otherwise wouldn't do.
01:20 darbelo kid51: It means people are going to go looking at a poorly formatted capy-paste that I don't have time to whack into shape right now. Not my intent at all.
01:21 kid51 Well, at least in principle, other people can improve the formatting.  It is a wiki, after all.
01:21 kid51 I used to correct a certain someone's spelling errors when he was writing a book on the wiki.
01:25 darbelo Well, if anyone wants to explain to google's wiki that those are pointers and not 'put stuff in bolfdace at random' markers I can hand out commit bits for the wiki.
01:26 * darbelo is getting into the spirit of user-editable documents.
01:26 cotto wikis are fun like that
01:27 * darbelo is also trying to sucker people into fixing his screwups while he goes to sleep.
01:27 Coke darbelo: {{{* this is a pointer}}}
01:28 kid51 afk
01:28 cotto darbelo, that's the spirit!
01:29 * Coke wonders if .const works on multis.
01:29 darbelo Yup. That works. Coke++
01:36 darbelo Coke: Oh wait, I meant the '{{{'. I have no idea about .const
01:39 particle joined #parrot
01:40 dalek decnum-dynpmcs: r169 | darbelo++ | wiki/DecNum.wiki:
01:40 dalek decnum-dynpmcs: Edited wiki page through web user interface.
01:40 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=169
01:45 dalek decnum-dynpmcs: r170 | darbelo++ | wiki/DecNum.wiki:
01:45 dalek decnum-dynpmcs: Opps. Hit 'save page' too soon.
01:45 dalek decnum-dynpmcs: A few more formatting adjustments.
01:45 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=170
01:46 Coke darbelo: btw, you can edit those with vi if it makes you happy.
01:46 darbelo Oh. No, thanks, vi doesn't make me happy at all :)
01:48 * darbelo uses Escape-Meta-Alt-Control-Shift.
01:48 Coke (the wiki is available next to /trunk as /wiki)
01:58 darbelo Well that's it for tonight. I'm going to catch some Z's before midnight catches me.
01:59 jrtayloriv joined #parrot
02:00 dalek partcl: r628 | coke++ | trunk/runtime/builtin/expr.pir:
02:00 dalek partcl: cleanup PIR
02:00 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=628
02:00 dalek partcl: r629 | coke++ | trunk/runtime/compilers.pir:
02:00 dalek partcl: expressions apparently need the \<NL> substitution run on them.
02:00 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=629
02:00 dalek partcl: r630 | coke++ | trunk/ (23 files):
02:00 dalek partcl: use more .const
02:00 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=630
02:00 dalek decnum-dynpmcs: r171 | darbelo++ | wiki/DecNum.wiki:
02:00 dalek decnum-dynpmcs: Trying out the svn checkout of the wiki. Coke++ for pointing it out.
02:00 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=171
02:01 darbelo left #parrot
02:02 Alias left #parrot
02:05 acajou left #parrot
02:27 Coke 1583 / 60
02:27 purl 26.3833333333333
02:28 Ryan52 1 / 0
02:29 Ryan52 :(
02:30 dalek partcl: r631 | coke++ | trunk/docs/spectest- (2 files):
02:30 dalek partcl: update spectest results.
02:30 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=631
02:34 Tene particle: I just noticed that the cell number I have for you is one digit too short. Can I get it from you again?
02:35 janus joined #parrot
03:01 jdv79 joined #parrot
03:01 jdv79 anyone know why cardinal's makefile calls rake?
03:04 bacek_at_work jdv79: because it's cardinal. And rake is "standard" ruby's building way.
03:05 jdv79 ok
03:06 jdv79 trying to build partcl and i get Can't open perl script "/usr/local/lib/parrot/1.5.0-de​vel/tools/dev/gen_makefile.pl"
03:06 jdv79 is that correct?
03:06 bacek_at_work Did you make install-dev from parrot build tree?
03:07 jdv79 no
03:07 bacek_at_work It will help :)
03:07 jdv79 i don't think rakudo or cardinal require that.  is this a new thing or a partcl only thing?
03:08 jdv79 i thought parrot_config took care of finding paths like that...
03:09 Austin That's the right path
03:09 bacek_at_work partcl can be built only from installed parrot.
03:09 Austin It's just not there.
03:09 bacek_at_work latest rakudo too
03:09 bacek_at_work So, you need install-dev to build them properly
03:10 Austin There used to be code in there somewhere that would check if languages were being developed under $PARROT/languages/foo. If you're using that model, a non-installed one might work. But I like installed. (Except for paying the update tax once a month. :(  )
03:13 jdv79 so there's no way to have multiple parrots around anymore?
03:15 jdv79 weird
03:20 donaldh joined #parrot
03:34 Andy joined #parrot
03:36 jdv79 thanks for the info
03:42 bacek_at_work jdv79: you can configure parrot with different prefixes.
03:45 jdv79 yeah, i didn't think partcl could handle that but it appears it can.  yay.
03:48 jdv79 http://pastebin.com/m50957fdd
03:48 jdv79 anyone get what that's all abount?
03:49 treed The cardinal makefile calls rake because the actual build system is rake.
03:49 treed But some people insist on using make still.
03:49 treed But the test harness exists in the Rakefile.
03:49 treed So in order to make test, you need to use rake.
03:49 jdv79 treed: it appears it does not emit tap.
03:50 treed What doesn't?
03:50 purl And I'm proud of it!
03:50 treed The test files certainly do.
03:50 jdv79 the cardinal make test
03:50 treed Why would it?
03:50 treed That's the harness.
03:50 treed It collects the data and reports.
03:50 jdv79 yeah, that's what i'm looking for
03:50 treed If you want tap directly, run the files directly.
03:50 jdv79 i guess i'll, yup, do a seperate harness
03:50 jdv79 cool, thanks
03:50 treed What is it exatly that you need?
03:51 treed The TAP output?
03:51 treed For what purpose?
03:51 jdv79 essentially i want to get cardinal status into something like smolder
03:51 treed Ah.
03:51 jdv79 and the cardinal default output seems like its lacking
03:51 jdv79 plus i'd have to write a custom parser i guess
03:51 treed I could make a new target for the rakefile if you tell me what format you need.
03:52 jdv79 just TAP
03:52 treed All you want is the all the TAP outputs catted together?
03:52 jdv79 but if your busy i'll end up scripting it eventually...
03:52 treed Nah, I have a few minutes right now.
03:53 jdv79 the test output needs to be seperated
03:53 treed Separated by what and into what?
03:53 jdv79 by the output of each test file
03:54 jdv79 preferrably files.  In the perl world we can just use TAP::Harness::Archive.
03:54 jdv79 i'm not entirely sure the easiest or best way to handle it here though
03:54 jdv79 no rush
03:54 treed So, say, you want the output of running t/00-sanity.t in t/00-sanity.tap?
03:55 jdv79 something like that
03:55 treed I can do that in a few minutes.
03:55 treed So if there's something more specific you need.
03:56 treed If you can give me an example of the format you'd prefer, I can try to match it.
03:56 Andy joined #parrot
03:57 jdv79 truth be told i was about to hit the sack - its midnight:)
03:57 treed Heh.
03:57 treed Does parrot or rakudo generate what you need?
03:57 jdv79 if you grab a smolder report file that's the ultimate format for now
03:57 jdv79 yes
03:57 treed I have the source for both of those.
03:57 TiMBuS joined #parrot
03:57 jdv79 parrot does
03:57 treed What make command should I run and what file(s) do I look at?
03:57 treed I can have it ready for you tomorrow.
03:58 jdv79 http://smolder.plusthree.com/app/​public_projects/tap_archive/26549
03:58 jdv79 that may be a good example
03:59 treed k
04:01 jdv79 make smolder_test in parrot makes that report
04:01 treed What's testfile there?
04:01 jdv79 hmm?  it runs over the whole suite i think.
04:01 treed It's just TAP output.
04:01 treed a single result
04:02 jdv79 i don't know that that is
04:02 jdv79 the stuff in t and meta.yaml are the important parts
04:02 treed k, I'll ignore it
04:02 treed not myconfig?
04:02 treed or config_lib.pasm?
04:02 jdv79 its not required for smolder
04:02 treed k
04:02 jdv79 that's extra
04:02 treed Pretty sure I can manage that.
04:03 treed the meta.yaml stuff may need some explanation.
04:03 treed But I can at least get things moving.
04:03 treed or I can look at parrot's makefile to see how it generates that stuff.
04:03 jdv79 ok, its probably generated by TAP::Harness::Archive.
04:04 treed How would it know what branch you're on and the version of Parrot?
04:04 jdv79 the extra_properties part of the meta file
04:05 treed Yeah.
04:05 treed that's what we're discussing
04:05 treed where that info comes from
04:05 jdv79 ah, that probably comes from custom code in parrot.
04:06 treed k
04:06 jdv79 lib/Parrot/Harness/Smoke.pm
04:07 jdv79 i think just the file_order list and the start and stop times are the only required parts of the meta
04:07 jdv79 anyway, no rush.  thanks.  goodnight:)
04:08 treed No problem.
05:10 hercynium joined #parrot
05:20 kyle_l5l joined #parrot
06:00 dukeleto joined #parrot
06:01 uniejo joined #parrot
06:10 dukeleto 'ello
06:17 dalek parrot: r40770 | dukeleto++ | trunk/docs/book/pir/ch04_variables.pod:
06:17 dalek parrot: [docs] Fix typo in ch04 of the PIR book
06:17 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40770/
06:17 cotto 'i
06:17 dalek lua: 47c85a4 | fperrad++ | src/pmc/luafunction.pmc:
06:17 dalek lua: use auto_attrs
06:17 dalek lua: review: http://github.com/fperrad/lua/commit/47​c85a4e98d2fc57726d543ce0f4ecfebe40b7e2
06:23 dalek parrot: r40771 | dukeleto++ | trunk/docs/book/pir/ch04_variables.pod:
06:23 dalek parrot: [docs] Add sinh to list of trig function in ch04 of the PIR book
06:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40771/
06:39 HG` joined #parrot
06:41 dalek parrot: r40772 | dukeleto++ | trunk/editor/README.pod:
06:41 dalek parrot: [docs] Update documentation in editor/README.pod regarding generating tags
06:41 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40772/
07:11 dalek tracwiki: v7 | dukeleto++ | BuildWarnings
07:11 dalek tracwiki: https://trac.parrot.org/parrot/wiki/Bu​ildWarnings?version=7&amp;action=diff
07:17 mokurai left #parrot
07:18 treed tell jdv79 that cardinal master's Rakefile will now produce such a tarball with "rake report". Let me know if you need anything different or anything added to it.
07:18 treed purl: tell jdv79 that cardinal master's Rakefile will now produce such a tarball with "rake report". Let me know if you need anything different or anything added to it.
07:18 purl treed: huh?
07:18 treed Hm
07:18 treed purl: msg jdv79 that cardinal master's Rakefile will now produce such a tarball with "rake report". Let me know if you need anything different or anything added to it.
07:18 purl Message for jdv79 stored.
07:18 treed there we go
07:20 dalek cardinal: eb932f5 | treed++ | Rakefile:
07:20 dalek cardinal: Add report target that generates a TAP archive.
07:20 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/eb932f564df452f23add8edc9274a1910d569db4
07:20 donaldh joined #parrot
07:29 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r40772 - Ubuntu 9.04 amd64 (gcc)
07:39 mikehh rakudo (9d9d416) builds on parrot r40772, make test / make spectest (up to 28061) PASS - Ubuntu 9.04 amd64 (gcc)
07:40 mikehh it's taking me just under 10 minutes to run make spectest in rakudo now :-}
07:42 dukeleto mikehh: what is your command line?
07:42 dalek parrot: r40773 | dukeleto++ | trunk/t/op/inf_nan.t:
07:42 dalek parrot: [t] Add TODO tests for fdiv+PMC+NaN bug
07:42 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40773/
07:43 mikehh just that - make spectest
07:44 dukeleto mikehh: do you have the shell variable TEST_JOBS set?
07:47 mikehh I am not running in a script - I think it is set for make spectest - I use in in the parrot tests
07:48 mikehh got to take my grandsons to school - bbl - it's less tanh 5 minutes away
07:51 dukeleto mikehh: tanh(5) = 0.999909204262595, so we will see you shortly :)
08:00 dalek parrot: r40774 | dukeleto++ | trunk/t/op/inf_nan.t:
08:00 dalek parrot: [t] Add TODO tests for cmod and mod, which have the same NaN bug as fdiv
08:00 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40774/
08:13 riffraff joined #parrot
08:25 mikehh dukeleto: :-} - had to take the dog out too
08:25 treed purl: msg jdv79 I've added some extra_properties as well. Since Cardinal doesn't have versions yet, I've substituted Commit, which is the git commit hash.
08:25 purl Message for jdv79 stored.
08:26 * treed heads to bed.
08:26 * purl follows treed
08:26 moritz just missed him :(
08:26 * dukeleto doesn't ask
08:26 mikehh I think this wireless keyboard sometimes does some strange things - like sending keys out of order or just not :-}
08:27 moritz purl: msg treed is there somewhere I could read about cardinal's progress? maybe a blog, or some test statistics like on http://rakudo.de/
08:27 purl Message for treed stored.
08:27 treed moritz: Would you like to read about its progress?
08:27 * treed isn't actually in bed yet.
08:27 treed There's nothing much yet.
08:27 treed I could start a blog for cardinal, or something.
08:28 treed we have a mailing list, but nothing much has been posted there yet
08:28 treed As for test statistics jdv79 is apparently looking into getting us on some kind of smolder site, which is the reason for the commits from tonight.
08:28 dukeleto treed: getting the word out about cardinal would be good.
08:29 * treed nods.
08:29 treed I'll look into starting a blog soon, I guess.
08:29 moritz treed: yes, I'd really like to read something
08:29 treed Taking suggestions for sites via purl-gram as I sleep. :-)
08:29 moritz and be sure to advertize the URL here, and get it onto the planet.parrotcode.org
08:29 * treed nods.
08:29 dalek cardinal: 7ccf0ae | treed++ | Rakefile:
08:29 dalek cardinal: Adds the architecture, branch, and platform extra_properties to the TAP archive.
08:30 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7ccf0ae44184250d2b17a2350eabdf08c1d7d523
08:30 dalek cardinal: 3be9b00 | treed++ | Rakefile:
08:30 dalek cardinal: Add extra_properties for submitter and commit (in lieu of version)
08:30 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/3be9b00320a97e170b27390359ed3ac34d2cb246
08:30 treed I did write up a report for a parrotsketch a few weeks ago at Tene's suggestion.
08:30 treed Unsure how many saw it/cared.
08:30 mikehh I've run some tests - I think I'll try now
08:32 moritz I saw it, yes
08:34 mikehh got undefined method `write' for nil:NilClass in make test
08:34 * treed blinks.
08:34 treed on master?
08:34 mikehh in cardinal tat is
08:34 treed What platform?
08:34 * treed hasn't seen anything like that.
08:34 treed Indeed, there is no such method.
08:34 mikehh Ubuntu 9.04 amd64 (gcc)
08:35 treed And I'm not aware that anything even attempts to use such a method.
08:35 treed Is the failure in the rakefile or in a test file?
08:36 treed Must be the rakefile
08:36 treed ack write indicates that the only calls to write as a method are there
08:36 mikehh hold a moment - it's picking the wrong parrot directory
08:36 treed Which line?
08:36 treed ah
08:36 treed It may do that.
08:36 treed the configurator isn't very well specified
08:36 treed How do you specify your parrot_config to other things? I can try to match that.
08:37 mikehh it's building in parrot but testing in g.parrot
08:37 treed If you're building with make, that'd be why.
08:37 * treed really should just dump the Makefile.
08:38 mikehh i did perl Conbfigure.pl --parrot-config=../parrot/parrot_config
08:38 treed I'm ambivalent on that, though, until I'm in a position to ensure that the rakefile works on Windows.
08:38 treed Yeah, the rakefile won't catch that.
08:38 treed It attempts to use whatever parrot_config is in path.
08:38 treed just a sec
08:39 mikehh I have about 5 or 6 parrot directories at any one time
08:39 treed Yeah.
08:39 treed I'm writing a way to override that detection.
08:41 mikehh how do you clean rake?
08:41 treed rake clean
08:42 treed if you want to get rid of all the end-binaries too, rake clobber
08:42 treed (which implies clean)
08:42 treed "rake -T" will give you a list of targets with descriptions
08:43 treed Okay, pull from master and that'll give you the ability to override parrot_config.
08:44 treed If build.yaml exists, make sure you delete it (rake clean should do it)
08:44 treed then set the PARROT_CONFIG environment variable to the proper path
08:44 treed (Once build.yaml is generated, that'll stick around.)
08:44 treed I tested it with:
08:45 treed PARROT_CONFIG=/Users/treed/local/bin/parrot_config rake config
08:45 treed You should see "Provided parrot_config reports..." rather than "Detected parrot_config reports..."
08:47 dalek cardinal: 5e1d506 | treed++ | Rakefile:
08:47 dalek cardinal: Provide method for overriding parrot_config detection via environment variable.
08:47 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/5e1d506f6257f94af7dbce55407207b0cf5e8840
08:48 mikehh my build.yaml is now ok but I still get undefined method `write' for nil:NilClass running rake test:all
08:48 treed What line?
08:49 treed Pretty sure that means it's having some trouble opening a file.
08:49 treed (ALthough why there's no exception...)
08:49 mikehh Rakefile:145:in `run_test'
08:50 treed Huh.
08:50 treed Oh, I know why.
08:50 treed Duh.
08:51 treed Pushed the fix.
08:52 mikehh ok - I'll try again
08:52 patspam joined #parrot
08:53 dalek cardinal: ad14323 | treed++ | Rakefile:
08:53 dalek cardinal: Some code for the report target wasn't properly guarded. treed--
08:53 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/ad14323e380a29c0b4c4e0167c862d272d32ef38
08:53 treed A full run of test:all takes about 9 minutes here.
08:53 treed Although I tried to optimize it for repeated runs.
08:53 treed If you don't change the compiler after the first run, it won't regenerate the PIR.
08:53 treed So successive runs only take about 30 seconds.
08:54 treed Really gotta fix the parser.
08:54 mikehh ok running
08:54 treed If all goes well, the last line will be "-- CLEAN FOR COMMIT --"
08:55 treed (I have a standing policy that no commit to master should ever lead to unexpected test failures.)
08:56 treed (Although the definitien of "unexpected" is a little elastic.)
08:56 treed (As long as there's an issue for it on the tracker and that's noted in the test file, it's not unexpected, even if it's new.)
08:58 mikehh looks good - The test suite took 3:30.
08:59 mikehh 603 tests were run, from 98 files.  519 tests passed, 0 of which were unexpected.  84 tests failed, 84 of which were expected.
09:00 treed wow, your system's much faster than mine
09:01 treed Anyway, it's now 2:00.
09:01 treed Really going to bed now.
09:01 treed Night!
09:02 mikehh Night
09:02 mikehh actually it's 10am for me
09:15 iblechbot joined #parrot
09:17 donaldh joined #parrot
09:24 mikehh bbiab
09:49 bacek joined #parrot
09:52 bacek o hai
10:05 MoC joined #parrot
10:35 bacek *incoming*
10:36 dalek parrot: r40775 | bacek++ | branches/context_pmc3/include/parrot/context.h:
10:36 dalek parrot: [core] Reindent Parrot_Context fields. No functional changes.
10:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40775/
10:36 dalek parrot: r40776 | bacek++ | branches/context_pmc3 (2 files):
10:36 dalek parrot: [core] Add lex_pad and outer_ctx accessors
10:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40776/
10:36 dalek parrot: r40777 | bacek++ | branches/context_pmc3/src (5 files):
10:36 dalek parrot: [core] Use Context.lex_pad and outer_lex accessor functions instead of poking in struct
10:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40777/
10:36 dalek parrot: r40778 | bacek++ | branches/context_pmc3 (2 files):
10:36 dalek parrot: [core] Remove unused Context.regs_mem_size field
10:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40778/
10:40 dalek parrot: r40779 | bacek++ | branches/context_pmc3 (2 files):
10:40 dalek parrot: [core] Remove unused Context.gc_mark field
10:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40779/
10:41 mikehh joined #parrot
10:43 dalek parrot: r40780 | NotFound++ | trunk/src/pmc/object.pmc:
10:43 dalek parrot: [cage] don't duplicate find override logic in Object.clone
10:43 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40780/
10:48 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40774 - Ubuntu 9.04 i386 (g++)
10:53 mikehh rakudo (9d9d416) builds on parrot r40774 - make test / make spectest (up to r28062) PASS - Ubuntu 9.04 i386 (g++)
11:15 mikehh joined #parrot
11:16 bacek joined #parrot
11:17 dalek parrot: r40781 | bacek++ | branches/context_pmc3/src/pmc (2 files):
11:17 dalek parrot: [core] Replace last Context.current_namespace poking with API function
11:17 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40781/
11:20 donaldh joined #parrot
11:20 dalek parrot: r40782 | bacek++ | branches/context_pmc3 (18 files):
11:20 dalek parrot: [core] Add current_cont accessor to Context and use it
11:20 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40782/
11:28 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40780 - Ubuntu 9.04 i386 (gcc)
11:35 mikehh rakudo (9d9d416) builds on parrot r40780 - make test / make spectest (up to r28062) PASS - Ubuntu 9.04 i386 (gcc)
11:35 Coke (partcl) did jdv79 get partcl working?
11:40 mikehh partcl: r631 builds on parrot r40780 make test - same results 6 test FAIL but all subtests PASS - Ubuntu 9.04 i386 (gcc)
11:40 polyglotbot OUTPUT[Parrot VM: Can't stat languages/tcl/tcl.pbc, code 2.␤main: Packfile loading failed␤]
11:41 mikehh shouldn't have put the colon there :-}
11:55 tokuhirom__ joined #parrot
11:58 dalek parrot: r40783 | NotFound++ | trunk/src/oo.c:
11:58 dalek parrot: [cage] Underscores? What underscores?
11:58 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40783/
12:00 mikehh cardinal (latest) builds on parrot r40780 - make test - 519 tests passed, 0 of which were unexpected,  84 tests failed, 84 of which were expected - Ubuntu 9.04 i386 (gcc)
12:00 bacek NotFound++ :)
12:01 mikehh switching back to amd64 - bbiab
12:03 NotFound Unfortunately I lost control of my amd64 machine, I updated and rebooted it forgetting that his grub was using by default another system without the net configured X-)
12:04 NotFound And I won't be at home until end of next week
12:04 bacek NotFound: it... happens
12:05 mikehh my main problem at the moment is getting a vm to work so I don't have to reboot - I only have a wireless connection
12:05 mikehh anyway I need to reboot now
12:05 NotFound I have an account on another amd64 machine, but guess where I have the ssh key for that account X-)
12:06 mikehh :-}
12:06 mikehh bbl
12:06 quek joined #parrot
12:09 MoC joined #parrot
12:11 szabgab joined #parrot
12:16 mikehh joined #parrot
12:18 mikehh back again (on amd64)
12:18 moritz opbots, trust mikehh
12:18 slavorg Ok
12:19 mikehh moritz: :-}
12:19 kid51 joined #parrot
12:19 moritz I didn't even know if that would work :-)
12:20 tokuhirom__ joined #parrot
12:20 Taulmarill joined #parrot
12:21 mikehh moritz: rakudo seems fine on both i386 and amd64 - still got to test NotFound's latest commit
12:22 NotFound mikehh: builds and test fine here
12:22 ruoso joined #parrot
12:24 mikehh partcl and cardinal seem ok - still got to get lua to build
12:26 moritz what's this dreaded commit everbody's talking about?
12:27 NotFound r40783?
12:27 mikehh not a problem - just haven't uped it yet
12:29 moritz ok
12:32 NotFound It should not have any problem, the underscore checking deleted is from internal tables, not from user provided names.
12:38 slavorgn joined #parrot
12:39 masak joined #parrot
12:47 dalek TT #937 closed by coke++: Testing new ticket creation.
12:49 * Coke wonders if any memory leaks have been plugged recently.
12:54 * bacek wave from context_pmc3 branch :)
12:58 * NotFound is almost ready to set auto_attrs in Object
13:04 mikehh bacek: how's context_pmc3 goin' - need any help testin'?
13:05 Coke NotFound: that should help memory pressure, neh?
13:05 Coke (help reduce..)
13:05 bacek mikehh: not yet. But it will require a lot of testing soonish
13:05 bacek mikehh: thanks for offer anyway.
13:05 mikehh ok - let me know
13:05 NotFound Coke: rakudo make test shows some speed improvement, but not noticeable.
13:06 Coke NotFound: k.
13:06 Coke I was disappointed to note that partcl is actually slower per-spec-test now that we're down to 40-something files than we were when we were running 60+. I can only hope that one of the spectests had a lot of short and fast tests that will bump our average time-per-test back down.
13:07 bacek mikehh: I definitely will :)
13:08 NotFound Coke: I hope that after context and pcc refactoring a lot of thing will be easier to diagnose and fix.
13:09 NotFound 1.6 will have much more clean internals
13:10 NotFound Maybe a good name for it might be "Mister Proper" X-)
13:11 quek left #parrot
13:12 bacek "Mister Almost Proper" :)
13:13 bacek oh noes... I hit some GC bug again in context_pmc3...
13:17 * Coke wonders again how to properly use the parrot call chain instead of rolling his own.
13:18 Coke Ah. I was going to let jonathan and pmichaud do it for perl6 first.
13:20 jonathan Coke: ...what are you waiting on us to do?
13:24 bacek jonathan: world domination of Perl6 which will render tcl totally obsolete?
13:24 * bacek hides
13:24 jonathan Ah, OK, sounds doable.
13:24 jonathan ;-)
13:24 * jonathan hides too
13:28 moritz anybody here wants to review https://trac.parrot.org/parrot/ticket/757 perhaps?
13:28 moritz the comments claim that the assembled patches fix the problem
13:28 moritz which would be a huge WIN for rakudo
13:28 moritz but I don't understand all that stuff
13:31 * bacek checking
13:31 Coke jonathan: perl6 lets you return further up the call chain than from where you were invoked, yes?
13:32 jonathan Coke: Ah, by invoking .leave on the block?
13:32 jonathan Coke: Yeah. However, we think we need some Parrot support for that...
13:33 jonathan IIRC that's what we concluded anyway.
13:33 NotFound rake is provided by ruby packages?
13:34 Coke jonathan: I need the same thing: [return -level 2]
13:35 bacek NotFound: bacek@icering:~$ dpkg -S `which rake1.9`
13:35 bacek ruby1.9: /usr/bin/rake1.9
13:36 Coke hurm. I wonder if that's where my insane memory usage is coming from.
13:37 particle anyone here use emacs?
13:37 particle i have a user asking for a .emacs snippet to set up his editor to use spaces for indenting rather than tabs
13:37 Andy joined #parrot
13:37 NotFound I had 1.8
13:38 particle actually, using perltidy for indenting will work, if anyone knows how to do that
13:38 bacek particle: apt-get install vim; vim; :set et
13:38 particle yeah, if only.
13:38 NotFound lua doesn't build with C++
13:39 Coke particle:
13:39 Coke http://tinyurl.com/l6xc5v
13:42 particle coke: i already sent a 'teach a man to fish' email, thanks :)
13:43 bacek "... If you teach a man to fish you can get rid of him for all weekends"? :)
13:45 nopaste "coke" at 72.228.52.192 pasted "bah." (10 lines) at http://nopaste.snit.ch/17667
13:47 Coke that pretty much chews up all memory.
13:47 rg particle: http://borkware.com/quickies/single?id=145
13:48 rg also, to get rid of tabs in the first place use untabify
13:56 Coke ah, it's worse than I thought:
13:56 Coke https://trac.parrot.org/parrot/ticket/945
13:57 Coke fixing that should take some pressure off the HLLs. :P
13:59 moritz ou.ch
14:05 bacek oh.shi
14:05 moritz do we have tests for this kind of stuff?
14:05 Coke for memory usage? probably not.
14:05 Coke Could probably use interpinfo to get a reasonable threshold for that sort of thing.
14:06 moritz that's something that could be testable with resonable effort, I think
14:06 Coke from PIR, even.
14:06 dalek parrot: r40784 | NotFound++ | trunk/src (3 files):
14:06 dalek parrot: [oo] set auto_attrs in Object PMC
14:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40784/
14:06 * Coke does a bisect.
14:07 dalek TT #945 created by coke++: Sub invocation consumes all memory.
14:07 * bacek must sleep
14:07 purl $bacek->sleep(8 * 3600);
14:07 bacek good night all
14:12 whoppix joined #parrot
14:13 Coke not surprisingly, I'm back in the range of "merge everything to trunk..."
14:15 Coke NotFound: I think it's you. =-)
14:19 pmichaud good morning, #parrot
14:19 Coke NotFound: yup. See #945.
14:22 Coke pmichaud: have you been having any memory issues in rakudo since 1.5.0 ?
14:22 Coke /since/post/
14:22 pmichaud Coke: I haven't been looking for them, so can't really say.
14:23 Coke k. I wasn't either, I just was using enough to start with that I went over the limit. =-)
14:25 Psyche^ joined #parrot
14:31 NotFound Coke: whitenight said he was checking for a possible memory leak, maybe is this.
14:35 * Coke wonders if he's relying on anything post that commit or if he can rollback partcl's dependency.
14:35 Coke hurm. is there an easy way to have git-svn update to a particular svn revision? or do I need to dig through the log and use the git key?
14:55 dalek parrot: r40785 | mikehh++ | trunk/src/pmc/class.pmc:
14:55 dalek parrot: codetest failure - fix tab in src/pmc/class.pmc
14:55 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40785/
15:06 mikehh All test PASS (pre/post-config, smoke, nqp_test, fulltest) at r40785 - Ubuntu 9.04 amd64 (g++)
15:11 dalek partcl: r632 | coke++ | wiki/ParrotIssues.wiki:
15:11 dalek partcl: Edited wiki page through web user interface.
15:11 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=632
15:11 dalek partcl: r633 | coke++ | wiki/ParrotIssues.wiki:
15:11 dalek partcl: remove old parrot bugs that no longer affect us.
15:11 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=633
15:14 Tene treed: if you start blogging about cardinal, make sure to get it added to planet.parrotcode.org
15:20 donaldh joined #parrot
15:21 whiteknight joined #parrot
15:22 mikehh rakudo (9d9d416) builds on parrot r40785 - make test / make spectest (up to r28062) PASS - Ubuntu 9.04 amd64 (g++)
15:25 mikehh messages
15:31 jhorwitz joined #parrot
15:33 mikehh partcl - r633 builds on parrot r40785 - make test - 6 tests FAIL but pass all subtests - Ubuntu 9.04 amd64 (g++)
15:34 mikehh the tests fail with attempt to access code outside of current code segment - 5 also report current instr.: '_main' pc 340 (src/tclsh.pir:166) - not t/cmd_lsort.
15:37 dalek parrot: r40786 | pmichaud++ | trunk/compilers/pge/PGE/Match.pir:
15:37 dalek parrot: [pge]  Fix .orig method on Match objects (RT #68680).
15:37 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40786/
15:37 dukeleto mikehh: that is what I was seeing with the tests as well
15:37 Coke jonathan: is it possible to override invoke from PIR or did you have to do it in C?
15:40 mikehh dukeleto: make spectest takes an hour or so so I don't run it every test run :-}
15:40 jonathan Coke: I wrote something in C to make it possible to do it in PIR.
15:41 Coke can that thing in C go into Sub.pmc? =-)
15:41 jonathan Coke: No, it needs to go into Object.pm, but it's a bit of a hack.
15:41 jonathan *.pmc
15:41 Coke hurm.
15:42 jonathan (We subclass Object in Rakudo.)
15:42 jonathan Let me get you a reference...
15:42 Coke I hesitate to cargo cult your p6 code into partcl; seems like this is something that should make its way back into core.
15:43 Coke (having nearly ripped out all the cargo culting from the original PerlInt stuff. =-)
15:43 jonathan http://github.com/rakudo/rakudo/blob/9d9d41675dab​91fbeb6dcf29998210e00baecf97/src/pmc/p6opaque.pmc
15:43 mikehh cardinal (current rev) builds on parrot r40785 - make test 519 tests passed, 0 of which were unexpected., 84 tests failed, 84 of which were expected. - Ubuntu 9.04 amd64 (g++)
15:43 jonathan Coke: If I put this hack into Parrot, I think somebody will yell at me.
15:43 Coke shorten that
15:43 purl That URL is at http://xrl.us/bfemuk [github.com]
15:43 jonathan And probably deservedly.
15:44 jonathan I think when the PCC re-writing lands, this will be fixable.
15:44 jonathan In a proper way.
15:50 pmichaud jonathan: you're more optimistic than I am here :-|
15:51 pmichaud but I think I'm just over-pessimistic
15:51 jonathan pmichaud: OK, I don't see why we'd not be able to fix it, in that we will be able to unshift the desried SELF onto the start of the CallSignature.
15:52 jonathan Rather than passing it on by cloning the invoke vtable method and setting a property to hack around the issue like Rakudo does now.
15:52 mikehh lua does not build with g++
15:52 jonathan It's entirely possible that allison will somehow thing that passing the args that would allow it to work is the wrong solution, however...
15:52 jonathan *think
15:52 pmichaud that's what I'm fearing :)
15:53 jonathan Well, we can just commit it, very quietly. ;-)
15:53 moritz :-)
15:53 Austin joined #parrot
15:55 jonathan pmichaud: BTW, I think I've just fixed a PCT bug.
15:55 pmichaud oh?
15:55 purl well, oh is that possible? better look into that...
15:55 jonathan pmichaud: You'll be unsurprised to know that it was in a bit of PCT that I wrote. :-P
15:55 jonathan We were using :pirop('setattribute'), but I'd not defined the signature for setattribute to say that the first element is an in, rather than a result.
15:56 jonathan Thus we generated very wrong code for $!a := 42
15:56 jonathan Of course, getattribute worked. ;-)
15:56 pmichaud ah, yes.
15:56 pmichaud I suspect setattribute should be    0PP
15:56 jonathan Missed an S, but yes.
15:56 pmichaud oh, right.
15:57 jonathan (for name)
15:57 jonathan Parrot tests pass, nqp tests pass, spectest'ing Rakudo now before I commit.
15:58 pmichaud are you also bumping Parrot REVISION?
15:58 jonathan oh wait, I shoulda svn up'd Parrot before doing this...
15:58 jonathan Yeah.
15:58 jonathan It best not be broken again.
15:58 pmichaud okay.  I'm in the middle of a parrot revision bump also.
15:58 jonathan Ah, OK
15:58 pmichaud I'll let you do the spectest and bump PARROT_REVISION.  Your bump will also fix RT #68680
15:58 pmichaud (the .orig method on Match objects)
15:58 jonathan Ah, nice.
15:58 * PerlJam boggles a little at git://github.com/mikecanz/gamebase.git
15:59 jonathan pmichaud: BTW, did you notice that Blizkost has sprung into existence? ;-)
15:59 pmichaud I'm going to grab lunch, then work on hague proposal, then see if I can eliminate string compares from 'isa' again.
15:59 pmichaud I did see Blizkost, yes.  Very cool.
15:59 jonathan OK, awesome.
16:00 jonathan Yeah, now it just needs people who actually know Perl 5 guts. ;-)
16:00 jonathan I'll keep pushing it along bit by bit, anyway.
16:01 pmichaud It's good to see it started.
16:01 pmichaud okay, I need lunch.  bbiaw
16:10 payload joined #parrot
16:15 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40786 - Ubuntu 9.04 amd64 (gcc)
16:19 whiteknight what is Blizkost?
16:21 jonathan whiteknight: http://github.com/jnthn/blizkost/tree/master
16:22 mikehh rakudo (9d9d416) builds on parrot r40786 - make test / make spectest (up to r28062) PASS - Ubuntu 9.04 amd64 (gcc)
16:23 whiteknight Well, that's pretty hot
16:23 whiteknight I thought at one point there was some interest in a full-on port of Perl5 to Parrot
16:24 jonathan whiteknight: There may still be, but that's rather more work.
16:24 jonathan It was tried a bit before, but Parrot just wasn't ready at that point.
16:25 whiteknight do you think it might be ready now?
16:25 * whiteknight really doesn't know what all Perl5 would require
16:26 jonathan whiteknight: I think it'd have a better chance of success now.
16:26 jonathan whiteknight: I think part of it was that Parrot was too much of a moving target.
16:26 whiteknight yeah
16:26 mokurai joined #parrot
16:26 moritz a full Perl 5 port to parrot wouldn't make XS modules work
16:27 whiteknight after chromatic's blog post the other day where he mentioned how p5 wasn't ported to any VMs like other languages are, I suspect tere might be some renewed interest
16:27 whiteknight moritz: no, that's true. But then again they would have access to extensions written in a more sane way
16:27 jonathan moritz: There is that issue too yes.
16:27 whiteknight XS isn't a "feature" so much as it is an "anchor"
16:27 jonathan I think there's probably room for both approaches to co-exist.
16:28 jonathan They probably have different characteristics.
16:28 jrtayloriv joined #parrot
16:29 whiteknight I imagine any port of p5 to any other system would remove some old "legacy" and "deprecated" thigns from it
16:29 jonathan whiteknight: Aye.
16:29 jonathan Anyway, we'll see how this approach goes.
16:29 treed mikehh++ # Adding cardinal to his testing suite
16:29 whiteknight jonathan: yes! I'm very excited about it
16:31 particle last time p5 on parrot was tried, parrot was much closer to the p5 vm
16:31 particle but parrot was also much less complete and correct
16:31 particle much of the steam from p5-on-parrot was taken away when pugs gained traction
16:32 particle before pugs, the perceived migration path to p6 was to swap out the vm first, then change the hll
16:32 particle after pugs, the perceived migration path changed, to swap out the hll (running on p5 as the vm), then swap out the vm
16:33 particle when pugs died, nobody cared about migration path anymore, since p5 and p6 communities had splintered
16:33 particle "perl 6 is not perl"
16:34 moritz which is kinda sad
16:34 particle yes, but that's been addressed
16:34 particle p5p and p6l are again speaking
16:35 particle e.g. dave mitchell consulted @larry about smartmatch semantics before releasing 5.10.1
16:35 rhr joined #parrot
16:35 moritz aye, and ben morrow asked p6l about various special blocks and exceptions
16:35 particle another good example.
16:35 purl i think another good example is Float <=> Num .   In Rakudo we have "Num" objects, but Parrot may occasionally throw us a "Float".  So we use the register method to try to remap Floats into Nums when we encounter them.
16:35 mikehh lua fail to build with gcc as well - it gets further though - error:imcc:make_pmc_const: no such pmc
16:35 particle purl, forget another good example
16:35 purl particle: I forgot another good example
16:36 Austin .
16:37 particle smop as p5 vm seems a likely migration path now
16:38 Austin left #parrot
16:38 mikehh cardinal (current) builds on parrot r40786 - make test - 519 tests passed, 0 of which were unexpected., 84 tests failed, 84 of which were expected. - Ubuntu 9.04 amd64 (gcc)
16:38 Austin joined #parrot
16:43 mikehh partcl - r633 builds on parrot r40786 - make test - same as before - Files=74, Tests=1321 - Ubuntu 9.04 amd64 (gcc)
16:52 Tene particle: can you send me your phone number again?  I have it in my phone with too few digits. :P
16:52 theory joined #parrot
17:09 pmichaud jonathan: no luck with your parrot revision bump?
17:10 jonathan pmichaud: I ran into some "fun"...seems some debug files left lying around that confused the linker, then it exploded.
17:10 jonathan :-/
17:10 jonathan spectest just finished
17:11 jonathan Looks fine, will commit my fix and bump.
17:13 jonathan pmichaud: done
17:13 pmichaud okay, great
17:13 pmichaud thanks
17:13 jonathan np
17:15 dalek parrot: r40787 | jonathan++ | trunk/compilers/pct/src/PAST/Compiler.pir:
17:15 dalek parrot: [pct] Fix code generation for binding to attributes.
17:15 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40787/
17:26 treed http://cardinalruby.blogspot.com/
17:26 treed One intro entry so far.
17:26 Coke treed: whole blog is cardinal related, or just things with certain tags?
17:28 pmichaud #ps in 62
17:29 treed The whole blog.
17:29 purl the whole blog is that way. a blog... for spam...
17:29 treed I have other places for other things.
17:29 Coke what name should show as the author?
17:29 Coke Treed? CodingRed ?
17:29 treed treed or Ted Reed
17:29 treed Coding Red is the title
17:30 Coke added.
17:30 treed Neat, thanks.
17:30 Coke that post should show up in the planet feed in a bit.
17:30 treed k
17:31 kyle_l5l joined #parrot
17:31 Coke 134610940 / 1024 / 1024
17:31 purl 128.374996185303
17:36 MoC So purl is checking every msg for a calculatable expression?
17:38 jonathan Either that or chance to make a snide remark. ;-)
17:38 Coke jonathan 2/3/4
17:38 iblechbot joined #parrot
17:38 Coke jonathan?
17:38 purl rumour has it jonathan is mailto:jnthn@jnthn.net or trying to put together a grant application. or however seeing weird issues
17:40 jonathan 42
17:43 Coke again, anyone looking for PIR-related hacking to do, there's a TON of stuff waiting for you in partcl. =-)
17:43 Coke Real knowledge of tcl not really necessary.
17:44 rdice joined #parrot
17:48 treed Actual content post on the blog now.
17:49 szbalint Coke: hm, I might take you up on that
17:49 szbalint I wanted to get familiar with profiling Parrot and using PIR
17:50 Coke szbalint: oh, I could DEFINITELY use help profiling. =-)
17:50 Coke partcl is sloooow. =-)
17:50 szbalint it's possibly simpler than rakudo too :P
17:51 dalek partcl: r634 | coke++ | wiki/SpecTestStatus.wiki:
17:51 dalek partcl: After downgrading our parrot to 1.5.0, we avoid the recent memory leaks
17:51 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=634
17:51 szbalint I could be wrong about that
17:51 dalek partcl: r635 | coke++ | trunk/config/PARROT_VERSION:
17:51 dalek partcl: bump down our parrot revision to avoid crippling memory leaks.
17:51 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=635
17:54 dalek parrot: r40788 | jonathan++ | trunk/compilers/pct/src/PAST/Compiler.pir:
17:54 dalek parrot: [pct] Second attempt at fixing binding of attributes. Must have messed up my test case before - this one really fixes it, it seems.
17:54 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40788/
17:55 Coke jonathan: do we have a parrot test for that?
17:55 Tene I can verify that you don't need to know TCL to work on partcl.
17:55 darbelo joined #parrot
17:55 Coke tene++ # helping dupe new commiters.
17:56 jonathan Coke: Apparently not. (yet)
17:57 allison joined #parrot
17:57 mikehh_ joined #parrot
17:57 Tene allison: in Seattle yet?
17:58 dalek partcl: r636 | coke++ | wiki/DevelopersGuide.wiki:
17:58 dalek partcl: for szbalint++
17:58 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=636
17:58 Topic for #parrotis now http://www.parrot.org | http://planet.parrotcode.org | 1.5.0 "TEH PARROTZ!"  Released!
17:58 allison Tene: yes, I'm here
17:58 allison Tene: you're teaching today?
17:58 Tene I am.
17:58 Tene Just taught the SSH chapter.
17:59 allison excellent! *nix training?
17:59 Tene Yes.
18:00 Coke szbalint: also, the easy bits of tcl are /really/ easy. =-)
18:00 Tene This is a custom class for Boeing.  All the students are from there.
18:00 Tene The lesson plan is "I dance and sing and do whatever they tell me, and default to this book if they don't speak up."
18:01 allison sounds like a great gig :)
18:01 Tene So what's your actual plan for tonight?
18:02 szbalint :)
18:03 Topic for #parrotis now http://www.parrot.org | Members, don't forget to vote today!
18:04 Coke allison++ # sending out the email reminder.
18:04 allison Tene: I'm headed over to particle's house for the 2pm meeting, then will be there until we all go for dinner
18:05 allison Tene: at least that's my current plan, it depends on what particle has arranged for tonight
18:07 kid51 joined #parrot
18:09 cotto once it's ready, would it be better to merge the profiling runcore before or after it knows how to use HLL annotations?
18:09 Coke before.
18:09 Coke annotations can be another branch.
18:09 Coke (and, "there are HLL annotations?"
18:09 Coke )
18:09 cotto that's what I thought
18:09 cotto lawl
18:09 Coke (as opposed to just annotations?)
18:10 allison cotto: since it's new, it's not breaking any existing features, so yeah, merge and then rebranch
18:10 cotto I thought annotations were primarily used for HLLs
18:10 cotto k
18:19 jhorwitz allison: ping
18:20 allison jhorwitz: pong
18:21 jhorwitz i have a commitment at 5:30 EDT (2:30 PDT) -- i assume the vote count, etc. won't last that long.
18:22 duk3leto allison: more info about the foundation elections?
18:23 allison jhorwitz: It shouldn't. And, there's no need to be around anyway, since you'll submit your vote electronically.
18:23 mokurai left #parrot
18:23 Coke allison; we should be able to subclass .Sub, yes?
18:24 allison Coke: yes (aside from known bugs with subclassing C PMCs from PIR)
18:25 cotto allison, src/tsq.c has no test coverage.  Does this mean that more tests need to hit it or that it needs to go away?
18:25 allison jhorwitz: I'll email the members list the results, and put together the unanimous board resolution appointing roles tomorrow
18:25 Coke k. I think I could speed things up in partcl quite a bit if I could have a hook that ran before the actual sub code.
18:25 allison cotto: it's not used any more, so can go
18:25 jhorwitz allison: ok
18:25 Coke (but I can't see way to do that currently that functions)
18:26 Coke allison: (board resolution) is it "same as last year"?
18:26 Coke (barring any surprises after initial voting)
18:26 allison cotto: (it used to be tested because it was a core part of concurrency scheduling)
18:27 allison Coke: aye, there's plenty of room to change roles if anyone wants to, but what we've got works well
18:27 cotto japhb, SHOUTING FILES made me lol
18:27 japhb :-)
18:28 cotto japhb, are you in contact with the toolchain hackers?  It seems like they had something to offer.
18:29 cotto (ooc)
18:29 bubaflub joined #parrot
18:29 japhb cotto, some, yes
18:30 duk3leto bubaflub: welcome
18:30 bubaflub righteous.
18:30 purl stick it to the man!
18:30 MoC joined #parrot
18:32 duk3leto bubaflub was my gsoc student this summer and is interested in working on parrot
18:32 bubaflub hey ya'll
18:32 Coke woot!
18:32 whiteknight what was his gsoc project?
18:32 treed purl has a case of stickittothemaneosis
18:32 purl treed: what?
18:32 treed Howdy bubaflub
18:32 bubaflub what's new?
18:32 whiteknight #ps in NAO?
18:32 purl new is the constructor method of Wx::App
18:32 Util NAO!
18:33 whiteknight yep, that's all "new" does
18:33 darbelo bubaflub: if you peek at http://irclog.perlgeek.de/parrotsketch/today you'll see the progress made by everyone in the past week
18:33 whiteknight meetings only on tuesday
18:35 darbelo Also, meeting in progress now :)
18:35 bubaflub so how can i help out with parrot? i'm pretty green
18:35 bubaflub not in the envy sense but in the newbie sense
18:36 darbelo bubaflub: You can do lots of stuff! What sort of stuff interests you?
18:36 Coke say tcl!
18:36 Coke =-)
18:37 bubaflub haha
18:37 bubaflub well, the summer of code stuff was some delicious math stuff
18:37 bubaflub i don't mind trying anything really
18:37 dalek rakudo: 8968778 | jonathan++ | src/builtins/control.pir:
18:37 bubaflub bug testing, implementing boring stuff
18:37 dalek rakudo: Unbreak eval of foreign HLLs.
18:37 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​9687785f351d1472492eaff5da87519d8c579c1
18:37 dalek rakudo: ab99068 | jonathan++ | build/PARROT_REVISION:
18:37 dalek rakudo: Bump Parrot revision to get fix in attribute binding code-gen, plus .orig fix from pmichaud++.
18:37 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​b99068d7f587f0da969b78dc44e4b015d8cc91f
18:37 MoC bubaflub: Maybe whiteknight++'s blog contains some useful hints: http://wknight8111.blogspot.co​m/search/label/Parrot4Newbies
18:38 bubaflub cool, will read
18:38 whiteknight I haven't written a Parrot4Newbies post in a while. I have to do that
18:41 Coke msg bacek : here's an iterator bug for you: http://rt.perl.org/rt3/Tic​ket/Display.html?id=40156
18:41 purl Message for bacek stored.
18:43 kurahaupo joined #parrot
18:47 Util Coke: (1) does Partcl still require an installed Parrot? (2) What Parrot rev should I use with Partcl until the memleak is fixed?
18:50 duk3leto whiteknight++'s blog posts in the Parrot4Newbies is some the best beginner info you can find about parrot.
18:50 hart joined #parrot
18:50 duk3leto bubaflub: you should create an account on trac.parrot.org
18:50 kid51 joined #parrot
18:50 cotto ttbot?
18:50 purl rumour has it ttbot is TapTinder build bot owned by mj41 and reporting http://tt.ro.vutbr.cz/buil​dstatus/pr-Parrot/rp-trunk build errors
18:50 dalek parrot: r40789 | pmichaud++ | trunk (3 files):
18:50 dalek parrot: [oo]  Remove the use of string name comparisons in VTABLE_isa_pmc checks.
18:50 dalek parrot: This results in a ~4.5% overall speed improvement for Rakudo's spectests.
18:50 dalek parrot: In the process we also eliminate the creation of bogus PMCProxy objects
18:51 dalek parrot: for instances of the Class PMC.  This commit also marks one test as "todo"
18:51 dalek parrot: that likely wasn't working properly to begin with.
18:51 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40789/
18:51 joeri joined #parrot
18:51 rhr joined #parrot
18:51 bubaflub duk3leto: did it.
18:52 bubaflub the only parrot on my system right now is the one that's built for rakudo
18:52 bubaflub should i pull in a separate dev copy?
18:52 duk3leto bubaflub: https://trac.parrot.org/pa​rrot/wiki/git-svn-tutorial
18:52 bubaflub mmmm delicious git-svn
18:53 duk3leto bubaflub: yummy yummy in my distributed tummy
18:54 hart g'day
18:54 duk3leto bubaflub: i suggest having a plain svn checkout of parrot as well, since certain test only get run in an svn checkout, like tests for svn-properties/etc
18:54 hart So I've read the intro pod on parrot and I'm still a bit lost as to what parrot actually is :-)
18:55 cotto parrot?
18:55 purl parrot is our teacher, our mother, our secret lover or the reason Dan started or the reason Dan left or pretty onionish:) or http://www.youtube.com/watch?v=​4vuW6tQ0218&amp;feature=related or regular releases and carefully following the deprecation policy
18:55 duk3leto pmichaud++ on the isa_pmc speedup
18:55 duk3leto hart: it is a virtual machine for dynamic languages
18:56 cotto fsvo "dynamic"
18:56 jonathan pmichaud: Nice - I guess a bunch of roles things didn't have an epic fail?
18:56 pmichaud jonathan: no fails.
18:57 jonathan pmichaud: Great. I musta failed before because of that mis-delegation.
18:57 hart ok so it's the bit behind perl, python and other dynamic languages
18:58 duk3leto hart: parrot is a bit lower level than perl/python. Perl/python could be built on top of parrot
18:58 hart understood - I get it now :)  Thanks.
18:58 duk3leto hart: many people are writing languages on top of parrot now, all are at different stages of alpha/beta :)
18:59 hart the impetus being standardisation?
19:00 cotto hart, one of them, yes
19:00 PerlJam hart: and clean perl innards
19:00 hart sounds good to me :)
19:00 cotto clean being a relative term
19:01 cotto it also exists to make life easier for people who want to write experimental languages without reinventing everything that it takes to create a modern language
19:02 darbelo Look arround, pick a thing different from the perl 5 internals. Congratulations! You have found something cleaner than the perl 5 internals!
19:02 PerlJam darbelo: heh
19:04 Coke Util: the version in config/PARROT_REVISION
19:04 Util Doh!
19:04 Coke util: yes, it does require an installed, and will continue to do so until parrot makes it easy to do both.
19:04 Util Coke: ok, thanks!
19:06 Coke Util: iwbni that parrot_revision had a way for me to say "no later than" in addition to "no earlier than". =-)
19:06 Util indeed
19:09 Coke it's just code. probably quite reasonable to do. (hint hint.) =-)
19:10 duk3leto jonathan++ on blizkost, looks freakin' cool
19:10 hart hehe
19:10 jonathan duk3leto: Want a commit bit? ;-)
19:10 hart thanks for the intro guys - tcn!
19:10 jonathan .oO( must recruit people to do the work for me )
19:10 hart left #parrot
19:10 jonathan ;-)
19:12 davidfetter joined #parrot
19:17 duk3leto jonathan: sure, i already forked it :)
19:21 jonathan duk3leto++
19:21 jonathan What's your github id?
19:22 jonathan duk3leto: ah, found you
19:26 dalek rakudo: 9efc9fe | pmichaud++ | build/PARROT_REVISION:
19:26 dalek rakudo: Bump PARROT_REVISION to take advantage of Class.isa_pmc improvements.
19:26 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/9​efc9fe5e005f218c3e6e363a09b49429d1355f0
19:27 duk3leto jonathan: my github is "leto"
19:34 NotFound whiteknight: have you any good idea about where to look for the memory leak?
19:34 whiteknight NotFound: I think it's in that function I pointed out to you the other day
19:34 jonathan duk3leto: done
19:35 whiteknight during the sweep phase, the GC is not deallocating the attribute structure
19:35 whiteknight it's not always calling VTABLE_destroy either, if the active_destroy flag is not st
19:35 whiteknight set
19:35 whiteknight fix that function, I think you'll save the world
19:36 NotFound whiteknight: In gc/api.c ?
19:36 whiteknight src/gc/gc_ms.c I think
19:36 NotFound No wonder I don't find it, is the only gc file I don't looked X-)
19:36 mikehh testing question - if a TODO test passes we get notified - how do we find out if a skipped test is now working?
19:37 whiteknight src/gc/gc_ms.c:free_pmc_in_pool
19:37 bubaflub duk3leto: i'm git-svn'in the repo
19:37 duk3leto mikehh: to my knowledge, skipped tests are not run. skipped tests are usually for when a feature is not there to be tested
19:37 duk3leto bubaflub: sweet, which rev are you starting from?
19:38 bubaflub well, the doc said 40000
19:38 bubaflub so now i'm waiting like a fool for not checking the current rev
19:38 duk3leto mikehh: but skips are used when for instance, parsing the test causes the test suite to blow up
19:38 bubaflub it's not that slow, actually, and i'm almost at 40500
19:38 mikehh duk3leto: a lot of test are marked skip because they caused a failure before - but do they still?
19:39 darbelo mikehh: un-skip'em and find out :)
19:40 mikehh that is not what I had in mind :-} rather running the test itself
19:41 Coke there is no way to say "run this skipped" test. TODOing is preferable unless it would kill the harness.
19:41 mikehh on its own - is there an easy way to do that?
19:41 Coke mikehh: not really.
19:42 mikehh ok I will just play around with it
19:45 NotFound whiteknight: looks like the world is safe :)
19:45 whiteknight that fixed it?
19:46 NotFound At least now the Sub test haven't eated even a 0.1% of memory in 2 minutes
19:46 darbelo cotto: ping
19:46 whiteknight awesome
19:46 whiteknight NotFound++
19:46 NotFound whiteknight++ hey, the karma is yours, you located it
19:47 Coke (memory) against head+patch, or 1.5.0+patch?
19:47 NotFound Coke: What patch?
19:47 purl We don't need no stinking patch!
19:47 Coke 15:45 <@whiteknight> that fixed it?
19:48 NotFound No patch yet, only my src
19:48 Coke yes, but are you testing against trunk, or against 1.5.0 ?
19:48 NotFound Gimme a few moments to do some testing.
19:48 whiteknight I told NotFound where to look, he fixed it
19:48 NotFound Trunk
19:48 Coke woot.
19:49 whiteknight I told you it would go quickly, I just haven't had a free moment to get in front of the editor
19:49 Coke is it posssible to dynamically alter a grammar after you compile it?
19:50 Coke (yet)
19:50 whiteknight how do you want to alter it? I think you can override protos
19:50 moritz Coke: for the optable parser that's possible, I don't think it works for ordinary grammars
19:50 Coke moritz: danke.
19:50 Coke I'm trying to puzzle a clean way to at least partially compile tcl.
19:51 Coke (as opposed to the full on runtime dispatch that means invoking the PIR compiler so much.
19:51 cotto darbelo, pong
19:52 Coke I was thinking that taking the initial builtins and making them part of the grammar (and then removing them from the grammar if they were overridden) would be a decent way to do it.
19:53 darbelo I checked out a copy of the pluggable_runcore branch and it faiuls to build on OpenBSD.
19:54 cotto darbelo, do you know how to fix it?
19:55 darbelo Sort of, the culprit is apparently a stray C_LIBS flags added to config/gen/makefiles/root.in
19:55 darbelo trunk has C_LIBS           := @libs@
19:56 darbelo branch has C_LIBS           := @libs@ -lrt
19:56 cotto yeah.  That's something that'll need to be fixed before the merge.
19:56 darbelo and my ld is choking on -lrt. But I don't know what that is supposed to be doing there.
19:57 cotto it's for clock_gettime et al
19:58 ttbot Parrot trunk/ r40773 i386-freebsd-64int make error http://tt.ro.vutbr.cz/file/cmdout/76847.txt ( http://tt.ro.vutbr.cz//buil​dstatus/pr-Parrot/rp-trunk/ )
19:59 mj41 ignore this ttbot report, I just kill make
20:01 darbelo cotto: Removing it fixes the build on OpenBSD amd64, I'll let you know how make test goes.
20:01 darbelo ping me if there's anything you need me to test, or something you want to try out.
20:01 purl I can't find me in the DNS.
20:05 darbelo All tests successful.
20:05 purl darbelo: that's because you wrote only one test, slacker!
20:06 * Coke wonders if it would be a speed improvement to have a ccache like hash for compiling PIR.
20:06 Coke (for partcl)
20:07 Coke (might be a win in loops)
20:09 cotto darbelo, does running a pir file with -Rprofiling work too?
20:10 darbelo Oh, silly me. Let me chack that.
20:14 darbelo Apparently it does, is there any output it should give me.
20:18 darbelo For example: The output of "parrot -Rprofiling examples/benchmarks/arriter.pir" is exactly the same as "parrot examples/benchmarks/arriter.pir" here. Should it be different?
20:19 Coke darbelo: look at parrot.pprof.
20:19 cotto no, but it should put a file in parrot.pprof
20:20 darbelo Oh, I see that.
20:20 darbelo F:examples/benchmarks/arriter.pir
20:20 darbelo S:parrot;arriter
20:20 darbelo CS:parrot;<unknown sub>@0xeb3c00
20:20 darbelo 27:0:new:0xeb3c00
20:21 darbelo and keeps going for another 2266946 lines
20:21 cotto can you nopaste the first couple dozen lines?  That second number shouldn't be 0.
20:22 nopaste "coke" at 72.228.52.192 pasted "super simplistic compiler cache." (77 lines) at http://nopaste.snit.ch/17670
20:22 Coke ... one wonders why I included MD5 in that second one.
20:25 nopaste "darbelo" at 200.49.154.172 pasted "output of parrot -Rprofiling examples/benchmarks/arriter.pir" (36 lines) at http://nopaste.snit.ch/17671
20:25 Coke darbelo: did you compile optimized?
20:25 cotto crap.  that's broken
20:26 dalek rakudo: 8b525f1 | jonathan++ | src/classes/Str.pir:
20:26 dalek rakudo: Catch exceptions in sprintf that come from Parrot, and just return a Failure.
20:26 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​b525f1c82fabdd03351708d71c2d927902af6eb
20:26 darbelo the 0 is the running time, right?
20:27 cotto darbelo, yes
20:27 darbelo what are you using to get it?
20:27 cotto clock_gettime
20:29 darbelo Hmm. Let me look into that.
20:38 treed Coke: Where did you add my blog?
20:39 Coke planet parrot feed.
20:39 Coke planet?
20:39 purl planet is a universal mystery, methinks.
20:39 Coke planet parrot?
20:39 purl i think planet parrot is an aggregate of any parrot-related blogs. or editable by coke or changable with a trac ticket.
20:39 treed Hm.
20:39 Coke planet parrot is also http://planet.parrotcode.org/
20:39 purl okay, Coke.
20:39 treed Haven't seen an update yet.
20:39 Coke treed: can take a while.
20:39 * treed nods.
20:40 Coke give it a day, I can ping robert/ask after that.
20:40 treed Wanted to make sure I was looking at the right place.
20:40 * treed nods.
20:41 * Coke poners removing Chip from the planet.
20:41 Coke "ponders"
20:41 treed Heh.
20:42 duk3leto how does one get on planet parrot?
20:42 Coke planet parrot?
20:42 purl planet parrot is an aggregate of any parrot-related blogs. or editable by coke or changable with a trac ticket. or http://planet.parrotcode.org/
20:42 moritz one asks Coke, it seems ;-)
20:42 * Coke also ponders removing leo.
20:43 Coke it helps to have a blog with at least one parrot post. =-)
20:44 moritz if the blog contains other posts, it's also nice to have a RSS feed or a parrot tag or category
20:45 treed moritz: http://cardinalruby.blogger.com/
20:45 treed As you suggested.
20:45 treed or blogspot.com I guess
20:45 moritz treed++
20:46 moritz Coke: can you also add http://cardinalruby.blogspot.com/ ? ;-)
20:47 treed He already has; just waiting for refresh.
20:47 particle coke: mail leo and see if he wants to stay on.
20:47 particle chip, too.
20:48 moritz ah, great
20:48 moritz sorry for the noise then ;-)
20:48 treed :-P
20:48 nopaste "darbelo" at 200.49.154.172 pasted "Big ass paste for cotto++, have fun!" (126 lines) at http://nopaste.snit.ch/17672
20:49 * treed also considers writing a short blurb on recent commits.
20:49 * treed has a phone interview in 10 minutes, however.
20:50 cotto darbelo, the problem is that -lrt is required on some platforms
20:50 cotto I guess I'll need a configure step after all.
20:50 darbelo cotto: The paste is meant as a sample of the parrot.pprof generated by running "parrot -Rprofiling examples/benchmarks/arriter.pir" with CLOCK_BEST #defined to a few different things.
20:51 darbelo cotto: Isn0t it easier to add/remove it on the hints file for the applopiate platforms?
20:51 cotto darbelo, you're right
20:51 cotto the only problem is that win32 will need a completely different set of functions
20:52 dalek parrot: r40790 | mikehh++ | trunk/src/oo.c:
20:52 dalek parrot: fix codetest failure - macros used in correct position - src/oo.c
20:52 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40790/
20:52 darbelo Does any of the "head parrot.pprof"s look sane to you?
20:52 cotto CLOCK_REALTIME looks best
20:53 cotto The nice thing about a configure step is that I could try several values and pick the best one, but I'd be glad to avoid that work.
20:53 darbelo You might want to make that hint-able too then,
20:53 darbelo have -DCLOCK_BEST=@clock_best@ in the flags passed to cpp
20:54 darbelo and then override/define @clock_best@ on the platforms that need it.
20:54 cotto also a good idea
20:58 hercynium joined #parrot
21:00 mikehh All tests PASS (pre/post-config, smoke, nqp_test, fulltest) at r40790 - Ubuntu 9.04 amd64 (g++)
21:04 jrtayloriv What would be the PIR equivalent of 'sweep 1' (from PASM)? Same thing? Or is that something you can't do from PIR?
21:06 jonathan same
21:08 jrtayloriv jonathan, thanks
21:09 mikehh rakudo (8b525f1) builds on parrot r40790 - make test / make spectest (up to r28067) PASS - Ubuntu 9.04 amd64 (g++)
21:09 bacek joined #parrot
21:16 mikehh partcl - r633 builds on parrot r40790 - make test - same as before - 6 tests FAIL, but all subtests PASS - Ubuntu 9.04 amd64 (g++)
21:18 mikehh cardinal (current) builds on parrot r40790 - make test - 519 tests passed, 0 of which were unexpected., 84 tests failed, 84 of which were expected. - Ubuntu 9.04 amd64 (g++)
21:19 mikehh ok - that's me for a spell - bbl
21:24 darbelo cotto: if you tell me what platforms need -lrt I can patch the hints files for you.
21:26 cotto so far I know that linux needs it
21:26 cotto I don't know about others
21:26 jrtayloriv Is there any difference between .end and .return if I don't pass any parameters to .return?
21:27 nopaste "darbelo" at 200.49.154.172 pasted "configuration patch for cotto++" (28 lines) at http://nopaste.snit.ch/17673
21:28 jrtayloriv Is .end just kind of like a closing } for the .sub directive?
21:28 moritz jrtayloriv: yes, you need it always (afaict)
21:29 jrtayloriv moritz, If I don't include .return, does the .sub return the value of the last line like Perl?
21:29 darbelo cotto: that nopaste should dtrt for you then.
21:29 moritz jrtayloriv: don't think so (but I'm not a PIR expert)
21:29 moritz jrtayloriv: but you can easily try out yourself
21:34 cotto darbelo, thank
21:34 cotto thanks, even
21:34 darbelo Hmm. Does anyone test parrot on solaris? Solaris might need "-lrt" to use clock_gettime.
21:34 cotto no idea
21:37 pmichaud if a .sub doesn't have a .return, then nothing gets returned
21:39 dalek parrot: r40791 | cotto++ | branches/pluggable_runcore/​config/init/hints/linux.pm:
21:39 dalek parrot: [configure] add hints to add -lrt to the libs on Linux courtesy of darbelo++
21:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40791/
21:42 jrtayloriv pmichaud, thank you.
21:44 bacek Good morning
21:44 purl And good moroning to you, bacek.
21:44 Limbic_Region joined #parrot
21:45 jrtayloriv I've been looking for some tedious busywork to occupy myself with while I am trying to grok the parrot internals. I came across whiteknight's blog and decided to take his advice and start converting PASM tests to PIR. Did I do everything correctly here? --> http://pastebin.com/m4e74494a
21:45 Whiteknight joined #parrot
21:49 joeri left #parrot
21:50 bacek Coke: can you create trac ticket for #40156 and assign it to me? I'm deep inside context_pmc3 branch now.
21:51 darbelo cotto: It looks like you left the -lrt in config/gen/makefiles/root.in
21:51 Whiteknight jrtayloriv++
21:52 bacek msg Coke: can you create trac ticket for #40156 and assign it to me? I'm deep inside context_pmc3 branch now.
21:52 purl Message for coke stored.
21:53 Whiteknight dukeleto: ping
21:53 darbelo build fails again after removing local changes and doing a svn up
21:54 bacek jrtayloriv: looks about allright
21:54 cotto darbelo, no I didn't. ;)
21:55 Whiteknight darbelo: how did your project end up?
21:55 hercynium joined #parrot
21:55 Whiteknight I've been trying to keep up, but haven't been able to the last few weeks
21:56 NotFound Whiteknight: a 'little' problem: the example program does not leak but rakudo doesn't build and partcl fails all tests
21:56 Whiteknight ...that is a "little problem"
21:56 Whiteknight nopaste the patch?
21:57 dalek parrot: r40792 | cotto++ | branches/pluggable_runcore/c​onfig/gen/makefiles/root.in:
21:57 dalek parrot: [config] remove hardcoded -lrt from root makefile
21:57 dalek parrot: cotto-- for not paying closer attention
21:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40792/
21:57 nopaste "NotFound" at 213.97.96.43 pasted "Attempt to fix memory leak" (23 lines) at http://nopaste.snit.ch/17674
21:57 darbelo Whiteknight: Eh, It does what it's supoposed to ('wrap decnumber'), but at the same time it doesn't ('provide useful big numbers')
21:58 Whiteknight darbelo: well that's still a good start! Is that the kind of project that new users could get involved with to help?
22:02 darbelo Whiteknight: Sure, there's still stuff to do there and I'll keep working on it. But I can't shake the feeling that killing decNumber and writting our own bignums would be better for what I originally had in mind there.
22:03 Whiteknight darbelo: Well, that's definitely an interesting idea. Whenever we use a library we are trapped into using it's semantics and limitations
22:04 Whiteknight of course, writing a functional and secure big number library is non-trivial
22:04 Whiteknight especially if we want performance to be better then lousy
22:07 darbelo Whiteknight: And it has to be plenty portable, too. But Parrot doesn't do too many esoteric things with it's bignums either. It's all basic arithmetic, log, ln, exp, pow and a few other functions.
22:08 Whiteknight so what's missing from the decnumber library that you wish we had?
22:08 Whiteknight arbitrary precision?
22:10 darbelo That's the catch. Right there, decNumber doesn't really provide arbitrary precision. After spending a lot of time looking at it I can't really point at anything inside the library and say "Parrot can't do that"
22:11 Whiteknight ok
22:12 Whiteknight we do currently have wrappers for GMP, is that any better?
22:12 Whiteknight I don't know alot of about the bignumber libraries
22:13 duk3leto Whiteknight: we currently have a very small subset of GMP wrapped. enough for the proper vtable interface, but there is a lot more in GMP that Parrot could have bindings for
22:13 darbelo I haven't really looked at them in depth, but gmp is a real arbitrary precision-lib, and damm fast too, so it's already looking better.
22:14 darbelo But, gmp is a pretty big dependency to pull in if all you want is to add two big integers together.
22:15 darbelo afk
22:15 Whiteknight okay
22:16 Whiteknight NotFound: Interesting, the backtraces I'm seeing seem to be producing an order-of-destruction error with the scheduler
22:16 dalek rakudo: bd51ce2 | jonathan++ | src/pmc/p6opaque.pmc:
22:16 dalek rakudo: Detect attempts to access attributes of a type object and throw an exception, rather than letting them be Null PMC Accesses.
22:16 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/b​d51ce26e3a76e1ce8eadc1e84a5060215de8846
22:16 duk3leto GMP *is* a big dependency, but that is not an issue, since it is an *optional* dependency, at least currently
22:17 bacek Whiteknight: yay! Looks like it is same problem with context_pmc3 branch. Exception handling is slightly broken.
22:17 Whiteknight NotFound: can we set attr_size for enum_class_Scheduler to 0, and manage it's memory manually?
22:17 Whiteknight and by "manage it's memory manually", I mean that we will just never free it
22:18 NotFound Whiteknight: let me do some test
22:18 Whiteknight I added a quick hack and will be testing that now
22:19 bacek *incoming*
22:21 Whiteknight hmm, that fix segfaults on miniparrot
22:21 Whiteknight same problem, although it shouldnt be happening
22:21 dalek parrot: r40793 | bacek++ | branches/context_pmc3 (15 files):
22:21 dalek parrot: [core] Add current_HLL accessor to Context and use it
22:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40793/
22:21 dalek parrot: r40794 | bacek++ | branches/context_pmc3/src/context.c:
22:21 dalek parrot: [cage] Fix docs of various Parrot_cx_* functions.
22:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40794/
22:22 * darbelo is back
22:24 Whiteknight NotFound: I think I have it
22:24 NotFound Whiteknight: good!
22:24 dalek parrot: r40795 | bacek++ | branches/context_pmc3 (7 files):
22:24 Whiteknight in src/pmc/scheduler.pmc:destroy, comment out the MUTEXT_DESTROY line
22:25 dalek parrot: [core] Add current_object accessor to Context and use it
22:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40795/
22:25 dalek parrot: r40796 | bacek++ | branches/context_pmc3/t/op/annotate.t:
22:25 dalek parrot: [t] Put pop_eh after catching exception in t/op/annotate.t
22:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40796/
22:25 dalek parrot: r40797 | bacek++ | branches/context_pmc3 (17 files):
22:25 dalek parrot: [core] Add current_sub accessor to Context and use it
22:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40797/
22:26 Whiteknight any git people in here?
22:26 * treed is kinda
22:26 * bacek looking around. Looks like no one here :)
22:26 Whiteknight how do I revert local changes?
22:26 treed git reset --hard
22:26 jonathan git reset --hard
22:26 bacek is it commited?
22:27 treed Oh, yeah, good question.
22:27 Whiteknight that did it, thanks!
22:27 Whiteknight tree++
22:27 bacek Bah! :)
22:27 Whiteknight jonathan++
22:27 treed git reset --hard works if it's not committed
22:27 darbelo karma tree
22:27 purl tree has karma of -2
22:27 jonathan If it is comitted I think you just panic.
22:27 Whiteknight bacek: are you kidding: I don't know how to commit either!
22:27 treed Otherwise you can do an in-place rebase.
22:27 jonathan ;-)
22:27 Whiteknight treed++
22:27 treed git rebase -i
22:27 purl somebody said git rebase -i was NOT a tool for viewing a branch
22:27 treed LOL
22:27 treed git rebase -i HEAD~5
22:28 treed that'll open a file in your editor with a list of the last 5 commits
22:28 treed Delete the lines for the commits you want to delete, then save and quit.
22:28 jonathan wow
22:28 treed You can also use that to squash commits into a single one, or reorder them.
22:28 treed Obviously, don't do this if you've already pushed somewhere.
22:28 bacek "git checkout filename" will work for single file
22:28 bacek "git rebase -i HEAD~2"; remove "pick <hash>" line;
22:28 treed (A good rule of thumb for any use of rebase.)
22:30 * bacek using git commit -i and git commit --amend quite often
22:33 NotFound Whiteknight: I give up for today, I'm tired
22:34 Whiteknight yeah, I know what you mean
22:34 Whiteknight I'll keep playing with it
22:36 bacek Whiteknight: Stop it! Put your hands above blanket! :)
22:36 jonathan !!
22:36 Whiteknight how embarassing!
22:37 Whiteknight NotFound: I'm running rakudo spectest now. Memory usage is pretty flat
22:37 cotto a response worthy or purl
22:37 duk3leto Whiteknight: "git checkout ." will act about the same as "svn revert -R" from the root of your checkout
22:37 rude_russian What?
22:37 Whiteknight so when this run is finished, I'll commit the fix if no problems and get more testing
22:38 duk3leto rude_russian: how can a lowly mortal help pirc?
22:38 NotFound Whiteknight: fine, I'll take a look at it tomorrow
22:38 duk3leto or bacek, even
22:38 Whiteknight purl I'll keep playing with it is <reply>Stop it! Keep your hands above the blanket!
22:38 purl OK, Whiteknight.
22:38 treed Is that like a white russian?
22:38 * treed needs to get some kahlua.
22:38 ruoso joined #parrot
22:38 jrtayloriv Whiteknight, Do you think that it would be better for me to create a separate patch for each test, or should I wait until I have the entire file converted and do them all at once?
22:38 * bacek passing kahlua and milk to treed
22:38 Whiteknight jrtayloriv: whatever you want, I'm happy to commit whatever
22:38 treed woot!
22:38 Whiteknight and more patches means more karma for you
22:38 duk3leto treed: I believe "caucasian" is the preferred nomenclature :)
22:38 * treed gets the wodka.
22:39 bacek duk3leto: better to ask kj
22:39 treed duk3leto: Only if you're The Dude. :-P
22:39 * duk3leto is The Dude
22:39 treed Well, shit howdy.
22:39 treed It's El Duderino.
22:41 rg1 joined #parrot
22:41 duk3leto treed: if you aren't into the whole brevity thing :)
22:42 dalek website: allison++ | 2009 Members' Meeting Minutes
22:42 dalek website: http://www.parrot.org/content​/2009-members-meeting-minutes
22:42 treed Which I'm not, as anyone who has to read my vociferous output could tell you.
22:50 dalek parrot: r40798 | cotto++ | branches/pluggable_runcore/​config/init/hints/linux.pm:
22:50 dalek parrot: [config] add word boundary assertions to code that looks at library flags
22:50 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40798/
22:52 darbelo You'd expect me, being the one that made the previous edit to that file and a user of an operating system that provide a 'rthread' library, to have noticed that one.
22:52 cotto particle, thanks for noticing that.
22:52 duk3leto ttbot ?
22:52 purl ttbot is TapTinder build bot owned by mj41 and reporting http://tt.ro.vutbr.cz/buil​dstatus/pr-Parrot/rp-trunk build errors.
22:54 duk3leto how do I add a node for taptinder?
22:55 bacek duk3leto: msg mj41
22:55 purl Sorry, I've never seen mj4 before.
22:56 darbelo cotto: I have some warning on the pluggable_runcore branch
22:56 darbelo src/runcore/cores.c:1177: warning: cast from pointer to integer of different size
22:56 darbelo src/runcore/cores.c:1182: warning: cast from pointer to integer of different size
22:56 darbelo src/runcore/cores.c:1189: warning: cast from pointer to integer of different size
22:56 darbelo src/runcore/cores.c:1189: warning: long long int format, HUGEINTVAL arg (arg 4)
22:57 darbelo Caused by the "(unsigned int) CONTEXT(interp)" casts in fprintf
22:57 duk3leto msg mj41 i am interested in helping out with ttbot, let me know if I can setup a node. I have access to some platforms that are currently not on there
22:57 purl Message for mj41 stored.
22:57 darbelo duk3leto: what platforms do you have access to?
23:00 duk3leto darbelo: mostly darwin-ppc, possibly others. I already have a "make smoke" cronjob that submits to smolder
23:00 darbelo does your systema have a librt.o ?
23:01 darbelo or .whatever-darwin-names-it's-shared-libraries for that matter.
23:01 duk3leto darbelo: at first glance, it seems not
23:02 darbelo good 'nuff for me. It's not something I'd expect in a BSD-derivative, but I'd hate to brake more platforms than I fixed with the -lrt patch.
23:03 darbelo I guess we'll find out for sure whne the branch merges :)
23:03 darbelo s/brake/break/
23:03 duk3leto darbelo: just let me know if you want testing on your branch before the merge. sometimes that helps :)
23:04 darbelo It's not mine. I just fixed the OpenBSD build of the pluggable_runcore branch.
23:05 darbelo But systems that provide a separate librt need to add a flag in their hints file.
23:07 darbelo Linux already has it and solaris might need one. I have no clue whatsoever about windows.
23:14 dalek parrot: r40799 | whiteknight++ | trunk/src (2 files):
23:14 dalek parrot: [gc] plug a big memory leak that was caused by the auto_attrs changes. The attributes structures weren't being freed in the GC. However, with the fix we were running into an order-of-destruction problem with the scheduler, so I commented out a line in the Scheduler PMC to avoid that
23:14 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40799/
23:17 dalek parrot: r40800 | whiteknight++ | trunk/src/pmc/scheduler.pmc:
23:17 dalek parrot: [gc] add a pointer to the associated TT for the line I commented out
23:17 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40800/
23:18 dalek TT #946 created by whiteknight++: Order-of-destruction error with Scheduler PMC
23:18 jonathan darbelo: What calls does librt provide?
23:19 Whiteknight I'm pretty interested in librt, but I don't know enough about it
23:20 darbelo On some systems you need it if you want to use clock_gettime and friends.
23:20 jonathan OK
23:20 * jonathan is curious how well the branch works on Windows.
23:20 jonathan Too tired tonight, but will try tomorrow or sometime.
23:21 darbelo libtr is the "POSIX.1b Realtime Extensions library"
23:21 darbelo librt is the "POSIX.1b Realtime Extensions library"
23:21 darbelo purl librt
23:21 purl somebody said librt was the "POSIX.1b Realtime Extensions library"
23:21 darbelo there.
23:21 darbelo purl forget libtr
23:21 purl darbelo: I forgot libtr
23:25 Limbic_Region jonathan - do you have a few minutes for me to regurgitate that really long FB msg that FB ate?
23:27 Austin joined #parrot
23:29 jonathan Limbic_Region: Yes.
23:29 jonathan Limbic_Region: Maybe in pm
23:29 Limbic_Region ok - yes, makes sense
23:30 jonathan Unless all of #parrot wants to know more about russian juicy girls.
23:30 * bacek is expert in this topic
23:31 bacek *incoming*
23:31 bacek $dayjob time, see you
23:31 dalek parrot: r40801 | bacek++ | branches/context_pmc3 (9 files):
23:31 dalek parrot: [core] Add current_pc accessor to Context and use it
23:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40801/
23:31 dalek parrot: r40802 | bacek++ | branches/context_pmc3 (9 files):
23:31 dalek parrot: [core] Add pred_offset accessor to Context and use it
23:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40802/
23:31 dalek parrot: r40803 | bacek++ | branches/context_pmc3 (9 files):
23:31 dalek parrot: [core] Add current_result accessor to Context and use it
23:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40803/
23:47 nopaste "darbelo" at 200.49.154.172 pasted "[PATCH] add \b to hint regexes, in the spirit of r40798" (123 lines) at http://nopaste.snit.ch/17675
23:48 Whiteknight purl msg bacek: Do you need anything from the context_pmc or context_pmc2 branches? If not we can delete them.
23:48 purl Message for bacek stored.
23:48 dalek parrot: r40804 | bacek++ | branches/context_pmc3 (24 files):
23:48 dalek parrot: Bring branch up-to-date with trunk.
23:48 dalek parrot: Conflicts:
23:48 dalek parrot: src/dynpmc/subproxy.pmc
23:48 dalek parrot: src/gc/mark_sweep.c
23:48 dalek parrot: src/pmc/addrregistry.pmc
23:48 dalek parrot: src/pmc/coroutine.pmc
23:48 dalek parrot: src/pmc/eval.pmc
23:48 dalek parrot: src/pmc/eventhandler.pmc
23:48 dalek parrot: src/pmc/exception.pmc
23:49 dalek parrot: src/pmc/hash.pmc
23:49 dalek parrot: src/pmc/lexinfo.pmc
23:49 dalek parrot: src/pmc/namespace.pmc
23:49 dalek parrot: src/pmc/nci.pmc
23:49 dalek parrot: src/pmc/object.pmc
23:49 dalek parrot: src/pmc/sub.pmc
23:49 dalek parrot: t/op/arithmetics.t
23:49 purl i think t/op/arithmetics.t is failing because 0.000000 ne -0.000000
23:49 dalek parrot: t/op/inf_nan.t
23:49 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40804/
23:50 darbelo purl forget t/op/arithmetics.t
23:50 purl darbelo: I forgot t/op/arithmetics.t
23:51 darbelo Anyone looking for easy karma? http://nopaste.snit.ch/17675 is your chance to get some!
23:52 kyle_l5l joined #parrot

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

Parrot | source cross referenced