Camelia, the Perl 6 bug

IRC log for #parrot, 2011-06-13

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:04 whiteknight benabik: yeah, it's non-working garbage
00:05 whiteknight I wanted to commit it to get more eyes on it, but the only eyes on it are for fixing the codestd issues
00:05 whiteknight so that's a failed experiment
00:09 kid51 is now known as kid51_at_dinner
00:10 plobsing ~~
00:10 whiteknight hello plobsing
00:12 plobsing where are you at with the pbc_merge stuff?
00:12 mikehh whiteknight: although I did some codetest fixes, I was also looking at the code, well trying to figgure it out anyway
00:12 whiteknight plobsing: I'm deleting what I have, and rebuilding to make sure I didn't screw anything up in the process
00:13 whiteknight it's become abundantly clear that I don't understand annotations enough to make this work
00:13 plobsing but you appeared to be on the right track
00:13 whiteknight I don't have nearly enough understanding of the details involved. I need to do a lot more code-reading before I can attempt this again
00:13 plobsing and you never told me what was going wrong. just describing it as "bad behaviour".
00:14 whiteknight plobsing: segfaults mostly.
00:14 plobsing ah
00:14 whiteknight lots and lots and lots of segfaults
00:14 plobsing in the merging or in the execution of merged code?
00:14 whiteknight mergine
00:14 whiteknight merging
00:15 whiteknight I'm going to play with some other code, figuring out how constants work and all. That's where I'm stumbling
00:16 plobsing ok then
00:17 whiteknight I was hoping I could shortcut, and just copy a bunch of code that works. This isn't that kind of problem
00:17 whiteknight sometimes there's no substitute for real learning and real researh
00:17 dalek parrot: a82fafc | Whiteknight++ | frontend/pbc_merge/main.c:
00:17 dalek parrot: remove my preliminary attempt at adding annotations to pbc_merge. I need to figure out the algorithm first before I put more code down
00:17 dalek parrot: review: https://github.com/parrot/parrot/commit/a82fafccd7
00:46 lichtkind left #parrot
00:59 dalek parrot/soh-cah-toa/hbdb: b0d9b5d | soh_cah_toa++ | include/parrot/hbdb.h:
00:59 dalek parrot/soh-cah-toa/hbdb: Reformatted some comments to be more consistent with the rest of Parrot
00:59 dalek parrot/soh-cah-toa/hbdb: review: https://github.com/parrot/parrot/commit/b0d9b5da74
01:04 dalek TT #2133 created by whiteknight++: src/extend.c Review
01:04 dalek TT #2133: http://trac.parrot.org/parrot/ticket/2133
01:07 whiteknight msg soh_cah_toa: Can you take a look at src/embed.c:Parrot_debug? Does it look like anything you are going to use, or anything worth saving?
01:07 aloha OK. I'll deliver the message.
01:07 soh_cah_toa whiteknight: i'll take a look now
01:07 whiteknight thanks. It's old and crufty, I just don't want to start killing something you want to keep
01:09 soh_cah_toa i am definitely all for nuking anything "old and crufty." if anything, i'd just copy the code to my project. let's see...
01:13 soh_cah_toa yeah, this can definitely be rewritten for hbdb
01:14 whiteknight okay
01:14 whiteknight i'm not going to delete it now because of deprecation, but it's on the chopping block
01:14 soh_cah_toa alright
01:15 whiteknight I'm not going to move it anyplace better, or save it, or document it or anything
01:17 soh_cah_toa btw, what exactly takes place during a deprecation? there's an old ticket about the whole "cwd in the search path" thing that i took care of but still needs deprecation. i'd like to take care of it but i'm not sure how it works
01:21 dalek parrot/leto/embed_grant: 53abb7d | dukeleto++ | t/src/embed/api.t:
01:21 dalek parrot/leto/embed_grant: Add coverage for Parrot_api_set_runcore and Parrot_api_(un)wrap_pointer
01:21 dalek parrot/leto/embed_grant: review: https://github.com/parrot/parrot/commit/53abb7dc74
01:36 whiteknight soh_cah_toa: there's a whole process, and we keep adding to it. there's a docs/project/support_policy.pod file that has most of the details
01:36 soh_cah_toa oh, ok. good
01:40 soh_cah_toa msg dukeleto could you send me the link to those old tests for parrot_debugger again? i thought i bookmarked it but i can't seem to find it anywhere
01:40 aloha OK. I'll deliver the message.
01:47 pjcj left #parrot
01:51 dalek parrot: 4ebc575 | Whiteknight++ | / (4 files):
01:51 dalek parrot: Move Parrot_new out of src/embed.c
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/4ebc575309
01:51 dalek parrot: 5593be7 | Whiteknight++ | src/embed.c:
01:51 dalek parrot: remove unused extern int Parrot_exec_run global
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/5593be7507
01:51 dalek parrot: 2af1cc7 | Whiteknight++ | / (4 files):
01:51 dalek parrot: move another function out of src/embed.c
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/2af1cc7911
01:51 dalek parrot: 37fe0b3 | Whiteknight++ | src/extend.c:
01:51 dalek parrot: update some docs in src/extend.c
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/37fe0b34bf
01:51 dalek parrot: 1fdeb68 | Whiteknight++ | / (5 files):
01:51 dalek parrot: Move flags, debug, and trace-related code from embed.c to src/interp/inter_misc.c
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/1fdeb68f0d
01:51 dalek parrot: 9fefc3a | Whiteknight++ | / (7 files):
01:51 dalek parrot: Parrot_pf_prepare_loaded_packfile -> Parrot_pf_prepare_packfile_init. Add a similar function for _load. Fix docs. Add some DEPRECATED: notices to embed.c with upgrade path information
01:51 dalek parrot: review: https://github.com/parrot/parrot/commit/9fefc3a698
01:51 whiteknight left #parrot
01:51 dalek TT #2134 created by whiteknight++: Move src/pbc_disassemble.c to frontend/pbc_disassemble/main.c
01:51 dalek TT #2134: http://trac.parrot.org/parrot/ticket/2134
01:51 dalek TT #2135 created by whiteknight++: Parrot_compile_file should take a compiler argument
01:51 dalek TT #2135: http://trac.parrot.org/parrot/ticket/2135
02:04 dukeleto msg soh_cah_toa git log -p -- t/tools/parrot_debugger.t
02:04 aloha OK. I'll deliver the message.
02:17 soh_cah_toa dukeleto: got it. thanks
02:24 soh_cah_toa no!
02:24 bubaflub joined #parrot
02:24 soh_cah_toa i was trying to make myself not auto-invisible
02:27 jsut joined #parrot
02:28 kid51_at_dinner Getting a lot of test failures in t/src/embed.t
02:28 kid51_at_dinner t/src/embed.t hangs
02:29 kid51_at_dinner is now known as kid51
02:29 kid51 Failed test 'call a parrot sub' at t/src/extend.t line 490 ; Exited with error code: [SIGNAL 11]
02:29 kid51 and so forth
02:30 kid51 left #parrot
02:32 jsut_ left #parrot
02:46 bubaflub left #parrot
03:05 theory joined #parrot
03:06 cotto ~~
03:07 soh_cah_toa cotto: hey
03:08 cotto hio soh_cah_toa
03:08 soh_cah_toa cotto: i took care of what we talked about earlier today
03:08 cotto I was just reviewing some commits and noticed that.
03:09 cotto soh_cah_toa: I think it's becoming increasingly important to figure out a testing strategy for the debugger (rather than just the C interface to the debugger).  Have you given it any thought?
03:10 soh_cah_toa ugh...
03:10 soh_cah_toa tests...
03:10 cotto yup
03:11 soh_cah_toa i'm not really sure where to start with tests
03:11 cotto The hard part will be writing the first one.  After that it'll get fun.
03:11 cotto then I'm glad I mentioned it
03:12 cotto I don't think it'll be too hard to run the debugger and mess with stdin, stdout and stderr via perl.
03:13 soh_cah_toa doesn't it need to be able to do something first? i'm not done working out the basic design (though i'm very close)
03:14 cotto soh_cah_toa: that's the beauty of tests.  You can write them first and use them to figure out how well a feature is doing.
03:14 cotto It's a joy to make a failing test pass.
03:14 soh_cah_toa oh yeah, i keep forgetting. test first
03:15 soh_cah_toa ok so once i have my basic design done and i'm just about to implement the first breakpoints...test
03:16 cotto you can write tests whenever.  I highly recommend trying to write them just after thinking of the design for the relevant feature, but there are other workable strategies.
03:17 soh_cah_toa right. did you read my most recent blog post?
03:17 cotto Part of their value is that they give you a chance to "use" your code before you write it, so you can start to get a feel for what the interface will be like.
03:17 cotto I didn't.  I was going to ask for now, but I won't now. ;)
03:17 cotto reading now
03:18 soh_cah_toa b/c i mentioned that once my basic design is done, i think the rest of the features to come (not just breakpoints) will come very very easily
03:18 cotto It's such a great feeling when you figure out what's been blocking you.
03:18 soh_cah_toa yes. it. is.
03:19 soh_cah_toa it has become so much more fun now that i'm not trying to do the impossible
03:19 cotto You don't need a whole lot of design just to get some very basic code working, but you've passed that point.
03:20 soh_cah_toa yeah
03:20 cotto good times
03:21 soh_cah_toa i know. now i actually look forward to the time i spend in front of my monitor all day :)
03:22 cotto What's your battle plan for the week?
03:22 cotto (it may not end up being what you do, but you should have a plan)
03:22 soh_cah_toa right
03:23 soh_cah_toa well, i should have a basic runloop done by tomorrow. after that i think i'll begin the tests for breakpoints
03:23 cotto using the pc?
03:23 soh_cah_toa yeah
03:24 soh_cah_toa that's just the way it's gonna have to be for now
03:24 cotto do the possible first
03:24 soh_cah_toa right
03:24 soh_cah_toa though i suppose that means i need a disassemble type command that can show the pc value of each opcode
03:25 cotto Yes.  That will vary between bytecode segments.
03:25 cotto pbc_dump -d will help
03:25 soh_cah_toa ah, yeah. i gotta review bytecode.pod
03:26 cotto Did you see my blog post on the topic?
03:27 soh_cah_toa no, but i'd love to
03:27 cotto http://reparrot.blogspot.com/2010/11/wh​at-happened-in-dynopmapping-branch.html
03:28 soh_cah_toa something that's always confused me is the difference between directories and segments. i think sometimes the terms are used synonymously and other times they're different
03:29 cotto you're probably right
03:29 cotto the docs need some love
03:29 soh_cah_toa definitely
03:30 cotto If you see something questionable, feel free to fix it in a branch and I (or someone) will verify that it's correct and merge it into master.
03:30 cotto or fix it in master
03:30 cotto several developers keep an eye on incoming commits
03:31 soh_cah_toa yeah, when i first joined, i noticed a lot of mistakes in the docs but didn't have enough experience or knowledge to correct them. now i think i can definitely clean some of them up
03:32 cotto great
03:42 cotto soh_cah_toa: how much perl have you written?
03:43 soh_cah_toa cotto: lots :)
03:43 cotto good.  That'll make writing tests easier.
03:43 cotto What have you written?
03:44 soh_cah_toa scripts. lots and lots of scripts. just things for typical system automation. a while ago i wrote one to tag, vbrfix, and apply a constant naming scheme to all my mp3's
03:45 soh_cah_toa i also started writing a script for automating releases but got caught up in gsoc stuff
03:45 soh_cah_toa parrot releases, that is
03:47 cotto soh_cah_toa: ok.  Have you worked much with cpan modules?
03:48 soh_cah_toa i've never written one myself. but i know about packages, Exporter, @ISA, yaddy ya...
03:49 bubaflub joined #parrot
03:50 soh_cah_toa i can install them as well
03:51 cotto soh_cah_toa: ok.  I want to check if you're used to using other people's code when available or if you're more used to rolling your own.
03:51 soh_cah_toa oh yeah. i've used plenty of cpan modules
03:57 cotto Great.  Other people's wheels usually  come pre-rounded.
03:58 soh_cah_toa ok. if Test::More is what you're getting at, i only just started looking at it a few months ago and have only written a few "hello world" tests.
03:58 soh_cah_toa i do understand how TAP works though. it's fairly simple. ok(), is(), isnt(), like(), blah, blah, blah
03:59 cotto soh_cah_toa: I'm asking more about general principles.  Test::More is delightfully easy to pick up if you know Perl.
04:01 soh_cah_toa yeah, TAP itself is quite simple. it's when and how to use it that i need to learn
04:02 soh_cah_toa though i know you're quite experienced w/ unit testing so once i get there, i'm sure if you just point me in the right direction i can get rolling alright
04:06 cotto Once you get the strategy and framework in place, the rest is straightforward.
04:07 soh_cah_toa exactly. that's what i expect
04:08 cotto In your place, I'd grab a couple pieces of scap paper and scribble out ideas until something looks feasible.
04:08 soh_cah_toa sure
04:19 hudnix left #parrot
04:20 woosley joined #parrot
04:46 bubaflub left #parrot
04:53 soh_cah_toa left #parrot
05:05 theory left #parrot
05:14 theory joined #parrot
05:14 theory left #parrot
05:16 cotto dukeleto: ping?
05:17 pmichaud joined #parrot
05:18 pmichaud parrot master is giving me lots of errors in "make test", and hangs at t/src/extend.t.  Known problem?
05:20 cotto It's news to me.
05:20 cotto looking into it now
05:21 JimmyZ joined #parrot
05:22 cotto I see a few build warnings and can verify that that test hangs
05:22 pmichaud I get lots of "Parrot VM: PANIC: Out of mem!"  errors also
05:23 cotto pmichaud: I don't see any of those.  Is that on your beefy machine?
05:23 pmichaud yes
05:24 cotto ah.  I see one of those in a test failure
05:30 cotto pmichaud: can you confirm that 9fefc3a69851ee662a5319f58498fdf047370463 (most recent commit in master) is the culprit?  It looks that way to me.
05:31 pmichaud http://gist.github.com/1022348  # list of failures I'm getting
05:31 pmichaud cotto: checking
05:34 pmichaud still building, checking
05:36 contingencyplan joined #parrot
05:37 pmichaud when I move to 1fdeb68f (i.e., HEAD^), the parrot build hangs at the compile of src/install_config.c
05:38 pmichaud shall I git bisect?
05:38 cotto pmichaud: sure
05:38 cotto I'm not installing anything, so I might be seeing different failures.
05:39 cotto I found at least one bug
05:40 pmichaud currently at RELEASE_3_4_0-154-g9b318f4
05:40 pmichaud (testing)
05:41 jsut_ joined #parrot
05:41 dalek parrot: 7477e65 | cotto++ | src/embed.c:
05:41 dalek parrot: pass a STRING* instead of a char* to Parrot_pf_read_pbc_file
05:41 dalek parrot: review: https://github.com/parrot/parrot/commit/7477e65c3f
05:41 cotto pmichaud: can you test with that commit?
05:41 pmichaud sure
05:41 pjcj joined #parrot
05:41 * pmichaud clones another repo
05:41 cotto I'm disappointed that that didn't produce a noisier warning.
05:41 Coke getting a ton of failures on master with SIGNAL 11
05:42 jsut left #parrot
05:42 pmichaud Coke: yes, I think that's what we're investigating now.   (btw, "hi!")
05:42 Coke pmichaud: yes, verifying your findings on OSX/86. ;)
05:42 pmichaud RELEASE_3_4_0-154-g9b318f4 fails
05:43 Coke "expected: nyarlathotep"
05:43 Coke ... that's just not right.
05:43 pmichaud oh, wait, maybe not.
05:44 cotto 7477 looks promising so far
05:44 Coke cotto:  t/src/extend_vtable.t
05:44 Coke still fails horribly with that commit.
05:44 pmichaud as a general note, parrot master "feels slower" to me somehow in its build than it did previously.
05:45 pmichaud I haven't benchmarked it yet but will do so tomorrow.
05:45 cotto Coke: I don't see any failures in that test.
05:45 cotto Coke: or do I?
05:46 cotto It seems to have failed under make test but not when I ran it by itself.
05:46 Coke perl t/harness <test name>
05:47 pmichaud 7477 passes for me, even under make test
05:47 pmichaud (the whole suite, not just extend_vtable...)
05:47 Coke commit 7477e65
05:47 Coke that fails for me.
05:49 cotto I have one test run that passes and one where extend_vtable fails
05:49 cotto wheee
05:51 cotto less broken is nice
05:54 cotto I'm not finding a reliable failure of extend_vtable.t
05:55 cotto It does seem to occasionally fail though
05:55 pmichaud 9fefc3a was reliably failing for me
05:55 pmichaud 7477e65 seems to be passing so far
06:02 cotto dukeleto's jitterbug seems to be stuck on 9fefc3
06:07 cotto I'm going to have to give up on extend_vtable.t until something reliable happens.
06:08 pmichaud git bisect claims that "1fdeb68f0d8ab4291b4387097a0a71bb654e632c is the first bad commit
06:08 cotto pmichaud: for which failure?  extend_vtable.t?
06:08 pmichaud for where I was getting multiple failures, including extend_vtable.t
06:14 pmichaud anyway, it seems to work now for me under 7477.  cotto++  (and thanks!)
06:15 cotto pmichaud: thanks for reporting it.
06:16 cotto I guess the automated tools got stuck on the hanging test.
06:16 cotto pmichaud++
06:41 cotto I'm putting together a document that I hope will make it easy to get a basic idea of what M0 is, why it's important and what it means for Parrot.
06:41 cotto Feedback is appreciated: https://gist.github.com/1022340
06:49 cotto msg dukeleto M0 glossy brochure draft at https://gist.github.com/1022340
06:49 aloha OK. I'll deliver the message.
06:54 Hunger left #parrot
06:55 allison cotto/lichtkind: the parrot logo is under this license http://wardley.org/computers/parrot/logo.html
06:57 cotto allison: did you make it?
06:57 allison cotto: I made the modified version that's on the current parrot.org site
06:57 cotto allison: what about the one on parrot.org?
06:57 allison but, it's based on Andy Wardley's original logo
06:58 cotto gotcha
06:58 allison different colors, and just the head
06:58 allison Andy said he'd be happy to assign the logo to PaFo
06:58 cotto allison: any luck finding papers relevant to M0 calling conventions?
06:58 allison so they could release it under any license
06:58 allison cotto: nothing relevant yet
07:00 cotto I wish the acm were either less useful or more open.
07:03 mj41 joined #parrot
07:04 allison cotto: yes, that would be nice
07:07 mj41 left #parrot
07:09 dalek parrot: a63a19c | mikehh++ | src/pmc.c:
07:09 dalek parrot: add missing ASSERT_ARGS
07:09 dalek parrot: review: https://github.com/parrot/parrot/commit/a63a19c430
07:10 cotto at least there's usually a preprint version floating around the tubes somewhere
07:11 cotto allison: do you have any thoughts on how to quickly figure out if a paper is (ir)relevant?
07:14 mj41 joined #parrot
07:21 mikehh embed changes have messed up tests with g++, ok on gcc
07:23 cotto mikehh: looking at it now
07:25 mikehh t/src/embed.t, t/src/extend_vtable.t and t/src/extend.t all fail with g++ build in make corevm/make coretest
07:25 fperrad joined #parrot
07:25 cotto mikehh: thanks.  I'm probably not in a good state to fix it atm
07:26 mikehh cotto: looking at it, will see what I can do
07:27 mikehh don't know how far I will get, other things are calling for my attention
07:28 cotto mikehh: finding reliable brokenness is great
07:30 cotto mikehh++
07:34 mj41 left #parrot
07:38 Hunger joined #parrot
07:38 mj41 joined #parrot
07:42 dalek parrot/m0-spec: 1921ceb | cotto++ | docs/pdds/draft/pdd32_m0.pod:
07:42 dalek parrot/m0-spec: rename iton to convert_n_i, etc
07:42 dalek parrot/m0-spec: review: https://github.com/parrot/parrot/commit/1921cebc09
08:05 mikehh All tests PASS (pre/post-config, make corevm/make coretest, make world/make test, fulltest) at 3_4_0-247-ga63a19c
08:05 mikehh Ubuntu 11.04 i386 (gcc)
08:06 mikehh but g++ is failing tests
08:06 cjh left #parrot
08:06 cjh joined #parrot
08:07 mikehh have to do something else now, will look at the g++ failures as soon as I can get back to it
08:21 particle left #parrot
08:24 particle joined #parrot
08:24 benabik left #parrot
08:40 ligne left #parrot
10:11 woosley left #parrot
10:59 ligne joined #parrot
11:31 contingencyplan left #parrot
11:34 ligne commit 4ebc5753093e29855a86211 seems to have caused g++ to explode a lot of its tests.
11:34 ligne t/src/embed.t:89:29: error: 'Parrot_new' was not declared in this scope
12:03 mtk joined #parrot
12:27 pmichaud http://gist.github.com/1022693   # parrot performance changes since 3.4.0
12:27 pmichaud http://gist.github.com/1022687  # and on a different machine
12:34 benabik joined #parrot
12:42 whiteknight joined #parrot
12:44 bubaflub joined #parrot
12:48 whiteknight good morning, #parrot
12:49 Coke cotto: the acm is completely useless because they are not open. discuss. ;)
12:50 Coke pmichaud: for https://gist.github.com/1022693, composite scores - is higher faster or slower?
12:51 atrodo Coke> What's to discuss?  That's truth.
12:56 ambs joined #parrot
13:02 pmichaud Coke: baseline (2011.01) is 100.  lower numbers indicate faster.
13:07 whiteknight hmmm, I wonder why master is running slower than 3.4?
13:09 whiteknight it's not much, but it is clear
13:10 Coke pmichaud: ah. then that's better than I thought. ;)
13:10 Coke pmichaud: how the hell are you?
13:10 Coke also, I think I'm supposed to bug you about a grant or something.
13:11 Coke also, pie!
13:11 Coke also also wik: http://news.perlfoundation.org/2011/0​6/2011q2-grant-proposal-port-pmt.html -- feedback from the community welcome.
13:11 pmichaud Coke: I'm doing okay.
13:12 pmichaud Grant is high on my mind -- indeed, it's the subject of my yapc::eu talk  (http://yapceurope.lv/ye2011/talk/3520)
13:12 pmichaud so, we should see progress/completion on that front soon
13:15 pmichaud (grant proposal):  yeah, I saw that proposal, but I'm not sure what to say about it yet.  My initial reaction was that it sounded like the proposer is unfamiliar with Perl 6's current module ecosystem (more)
13:16 pmichaud on the other hand, having someone put together a chain of tools like this might really push development a bit
13:16 Coke My first thought was "what are pmtools?" ;)
13:16 pmichaud I had that thought also :)
13:17 atrodo Me too.  I thought it was a set of perl tools by pmichaud
13:17 pmichaud lol
13:17 Coke atrodo: ditto!
13:18 pmichaud some of my students at the university thought I was some uber-Perl-5-programmer because they looked in /usr/lib/perl5 and saw a ton of ".pm" files.
13:18 tadzik :D
13:19 Coke right conclusion, bad premise.
13:24 * atrodo will now attribute to the existence of afternoon to pmichaud
13:24 PerlJam good morning.
13:25 pmichaud I'm not responsible for it's existence... it was just given that abbreviation in my honor.  :-)
13:26 pmichaud *its
13:26 pmichaud anyone have any recommendations for a new wireless keyboard?  the one I have now is becoming suboptimal
13:26 PerlJam Coke: re pmtools proposal ... it actually made me sad because it seems out-of-whack with reality.  Now, had he proposed to recreate the Perl Power Tools in Perl 6 .... *that* I might be able to support  :)
13:33 Coke commenting on the article will help.
13:40 benabik G'morning, #parrot
13:40 whiteknight left #parrot
13:43 mikehh pmichaud: I recently got a MS Desktop 5000, which has worked very well for me
13:44 pmichaud mikehh: thanks
13:44 pmichaud Coke: I'll add a comment to the article later tonight if I can
13:44 mikehh I am using it on Ubuntu/Kubuntu and have had no problems with it pt the mouse
13:45 hudnix joined #parrot
13:45 plobsing left #parrot
13:45 mikehh s/pt/or/ - well sometimes it takes a little bit of time to get used to the key spacings
13:47 * JimmyZ commented to that grant
13:49 Coke pmichaud, perljam: thanks for the feedback.
14:02 Coke we have some really slow tests.
14:03 Coke t/compilers/opsc/08-emitter.t t/compilers/opsc/06-opsfile.t t/src/extend_vtable.t t/compilers/opsc/02-parse-all-ops.t
14:03 Coke (8-13s)
14:14 contingencyplan joined #parrot
14:14 PacoLinux joined #parrot
14:16 estrabd left #parrot
14:32 RobertLJ joined #parrot
14:33 RobertLJ left #parrot
14:36 RobertLJ joined #parrot
14:37 bluescreen joined #parrot
14:38 Coke yay, partcl-old still passing all tests.
14:38 Coke partcl new still failing the same few tests. :(
14:39 RobertLJ left #parrot
14:39 RobertLJ joined #parrot
14:41 JimmyZ left #parrot
14:46 dalek parrot/nqp_pct: 308137b | benabik++ | compilers/pct/src/PAST/Compiler.pm:
14:46 dalek parrot/nqp_pct: War on Q:PIR - PAST::Compiler.if
14:46 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/308137bf80
14:46 dalek parrot/nqp_pct: 53565cf | benabik++ | compilers/pct/src/PAST/Compiler.pm:
14:46 dalek parrot/nqp_pct: PAST::Compiler.wrap_handlers fixes
14:46 dalek parrot/nqp_pct:
14:46 dalek parrot/nqp_pct: Fix errors left by benabik-- when he forgot to run `make test`.
14:46 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/53565cf42b
15:15 klavs joined #parrot
15:19 mtk left #parrot
15:24 mtk joined #parrot
15:32 klavs left #parrot
15:53 dalek parrot/nqp_pct: 62c7096 | benabik++ | compilers/pct/src/PAST/Compiler.pm:
15:53 dalek parrot/nqp_pct: [pct]: 'cconst' should be 'cclass'  (there is no 'cconst.pasm').
15:53 dalek parrot/nqp_pct:
15:53 dalek parrot/nqp_pct: Originally f2c7bafe48bdb651b3bae9a2a3c07f00d59529f8
15:53 dalek parrot/nqp_pct: Author:pmichaud <pmichaud@pobox.com>
15:53 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/62c7096fe9
15:53 dalek parrot/nqp_pct: 832e3eb | benabik++ | compilers/pct/src/PAST/Compiler.pm:
15:53 dalek parrot/nqp_pct: [pct]: Add a '!macro_const' type to PAST::Val
15:53 dalek parrot/nqp_pct:
15:53 dalek parrot/nqp_pct: This allows any value to be used as a pasm constant.  (Note that the
15:53 dalek parrot/nqp_pct: caller is responsible for providing any .include statements needed in
15:53 dalek parrot/nqp_pct: this case.)
15:53 dalek parrot/nqp_pct:
15:53 dalek parrot/nqp_pct: Origionally 197f87889f4154e4df4e7c3f2c680d39324b881d
15:53 dalek parrot/nqp_pct: Author:pmichaud <pmichaud@pobox.com>
15:53 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/832e3ebc42
15:53 dalek parrot/nqp_pct: c4d0cf0 | benabik++ | compilers/pct/src/PAST/Compiler.pm:
15:53 dalek parrot/nqp_pct: [pct]:  Improve PAST::Val const ability to remain constants in PIR.
15:53 dalek parrot/nqp_pct:
15:53 dalek parrot/nqp_pct: Origionally 9ebd8452c63fe66970a1248ac328d137625406c8
15:53 dalek parrot/nqp_pct: Author:pmichaud <pmichaud@pobox.com>
15:54 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/c4d0cf0104
15:54 dalek parrot/nqp_pct: 5587459 | benabik++ | compilers/pct/src/PAST/ (2 files):
15:54 dalek parrot/nqp_pct: [pct]:  Add a "childorder" property to PAST::Node
15:54 dalek parrot/nqp_pct:
15:54 dalek parrot/nqp_pct: Allows child nodes to be evaluated in reverse sequence.
15:54 dalek parrot/nqp_pct:
15:54 dalek parrot/nqp_pct: Originally 5d77384daf090fd361e913abc5e07b7e70426bd1
15:54 dalek parrot/nqp_pct: Author:pmichaud <pmichaud@pobox.com>
15:54 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/5587459a9a
15:54 dalek parrot/nqp_pct: a71400c | benabik++ | compilers/pct/src/P (22 files):
15:54 dalek parrot/nqp_pct: Re-bootstrap PCT
15:54 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/a71400c309
15:54 klavspr joined #parrot
15:55 dalek parrot/nqp_pct: 5a9b56d | benabik++ | / (62 files):
15:55 dalek parrot/nqp_pct: Merge branch 'master' into nqp_pct
15:55 dalek parrot/nqp_pct:
15:55 dalek parrot/nqp_pct: Conflicts:
15:55 dalek parrot/nqp_pct: compilers/pct/src/PAST/Compiler.pir
15:55 dalek parrot/nqp_pct: compilers/pct/src/PAST/Node.pir
15:55 dalek parrot/nqp_pct: review: https://github.com/parrot/parrot/commit/5a9b56d499
15:56 pmichaud benabik: I have another big batch of updates coming in shortly
15:56 benabik pmichaud: Sounds like fun.  :-)
15:56 pmichaud just wanted to give you a "heads up"  :)
15:57 benabik pmichaud: The register changes for Rakudo?
15:57 pmichaud yes, for better management of temporaries.
16:01 Coke but they're only temporary!
16:01 Coke ;)
16:05 dmalcolm joined #parrot
16:08 JimmyZ joined #parrot
16:09 lucian joined #parrot
16:11 lucian good day
16:13 JimmyZ left #parrot
16:14 JimmyZ joined #parrot
16:17 Coke lucian: hio
16:21 lucian Coke: saying 'good morning' at this sort of time bothers me :)
16:24 Coke "close enough"
16:24 PerlJam lucian: good morning!  ;>
16:27 theory joined #parrot
16:41 lucian PerlJam: :P
16:43 * Coke runs make test on rakudo for the first time in ages.
16:47 cotto_work ~~
16:50 mj41 left #parrot
16:53 mikehh hi cotto_work
16:55 NotFound ~~
16:56 mikehh t/src/embed.t, t/src/extend_vtable.t and t/src/extend.t all fail with g++ due I think to Parrtot_new bring moved to interpreter.h but \i can't seem to fix it
16:56 davidfetter joined #parrot
16:57 mikehh tried various things, probablymissing something obvious
16:58 cotto_work stuff's on fire at $dayjob.  will look when I can
16:58 mikehh the tests pass with a gcc build
16:59 mikehh all the tests fail with Parrot_new not declared in scope
17:04 dalek parrot: 10610ab | mikehh++ | src/extend.c:
17:04 dalek parrot: remove another compiler warning ["/*" within comment]
17:04 dalek parrot: review: https://github.com/parrot/parrot/commit/10610ab147
17:05 mikehh not altogether happy about that one, but it did remove the warning
17:17 JimmyZ left #parrot
17:27 dukeleto ~~
17:28 cotto_work hio dukeleto
17:28 dukeleto Smolder is still down. We need somebody to fix that.
17:28 dukeleto cotto_work: mornin'
17:31 lucian left #parrot
17:35 dalek parrot: e2857ba | dukeleto++ | t/src/embed/api.t:
17:35 dalek parrot: Merge branch 'leto/embed_grant'
17:35 dalek parrot: review: https://github.com/parrot/parrot/commit/e2857ba7af
17:36 Coke dukeleto: isn't OSU on the smolder issue?
17:38 lucian joined #parrot
17:40 hudnix left #parrot
17:42 whiteknight joined #parrot
17:42 RobertLJ left #parrot
17:42 cotto_work dukeleto: any thoughts on the M0 glossy?
17:49 lichtkind joined #parrot
17:49 lichtkind the "new" GC is full operational?
17:50 whiteknight which is the "new" one, genms?
17:51 cotto_work gms is the newest and is the current default
18:03 lg_quassel joined #parrot
18:04 cotto_work whiteknight: I'm writing an introduction to M0 and would appreciate your feedback: https://gist.github.com/1022340
18:04 whiteknight i'll take a gander at it tonight
18:05 cotto_work thanks
18:05 losinggeneration left #parrot
18:14 dukeleto cotto_work: i like the m0 brochure, nicely done
18:25 cotto_work dukeleto: anything that could use improvement?
18:25 cotto_work (apart from formatting, links, etc)
18:26 mj41 joined #parrot
18:26 cotto_work also, what do you think would be the best format?
18:28 dukeleto cotto_work: the text is great. But it is missing the "glossy" part of the brochure ;)
18:28 dukeleto cotto_work: we need some kind of graphics
18:28 dukeleto cotto_work: i am thinking Markdown would be good. We can include an image via markdown and have it sit on github
18:30 cotto_work dukeleto: that's not a bad idea.  I want it to be as accessible as possible.
18:32 contingencyplan left #parrot
18:33 benabik Hm.  extend_vtable.t is segfaulting left and right.
18:34 cotto_work benabik: still?  I fixed something last night.
18:34 benabik cotto_work: Yes, at least on OS X/Intel.  I'm bisecting it now.  :-)
18:36 cotto_work benabik++
18:36 benabik I have no idea how people debugged things before git-bisect.
18:37 cotto_work banging diffs together and flipping bits with a very tiny magnet and a pair of tweezers
18:38 benabik :-D
18:39 klavspr left #parrot
18:39 atrodo You had tweezers?  Back in my youth... Oh, wait, I had floppies
18:40 dukeleto benabik: more segfaults? Do you have the prove -v output ?
18:40 klavs joined #parrot
18:40 dukeleto benabik: i can bisect in my head for that file :)
18:40 benabik dukeleto: 140 ERROR 11s.
18:40 tewk left #parrot
18:40 benabik dukeleto: The original output is buried in backscroll of bisects by now.
18:40 dukeleto benabik: every test coredumps?
18:40 * dukeleto feels proud
18:41 cotto_work I see a couple failures in t/src/extend.t
18:42 cotto_work dukeleto++
18:43 * dukeleto has been using g++ and all tests pass for me. This stuff sounds platform-dependent
18:44 benabik Hm.  bisecting would probably go faster if I just used prove extend_vtable instead of the full `make test`
18:54 dukeleto benabik: indeed.
18:54 dukeleto benabik: prove -v t/src/extend*.t
18:54 pmichaud there was some reports last night that 'prove' would pass while 'make test' would fail
18:55 benabik No, prove is failing in the same spectacular manner.  :-)
18:55 pmichaud okay.  :)
18:55 pmichaud http://irclog.perlgeek.de/p​arrot/2011-06-13#i_3915618  # reference in case anyone wants to look
18:55 benabik 4ebc575 (Move Parrot_new out of src/embed.c) is the first bad commit
18:56 whiteknight FFFFFUUUUUUU
18:56 benabik Probably is a difference in the way Linux ld.so and OS X dyld handle things.
18:56 whiteknight what is the test saying? I didn't change the functionality of that function, I only moved it to a different file
18:56 benabik whiteknight: 140x SIGNAL 11s.
18:57 * whiteknight hates
18:57 whiteknight see whiteknight hate. Hate, whiteknight. Hate.
18:57 benabik Isn't cross-platform compatibility fun?
18:58 pmichaud it might not be entirely platform-dependent, since I was having trouble last night on Ubuntu 64-bit
18:58 ligne no segfaults here (linux x64), but a lot of whining about how "‘Parrot_new’ was not declared in this scope"
18:59 whiteknight those tests don't include parrot/parrot.h
19:00 whiteknight TT #2124
19:02 whiteknight Can somebody add "#include "parrot/parrot.h"" to include/parrot/extend.h?
19:02 benabik Adding parrot/parrot.h to extend_vtable.t gets it working again.
19:02 whiteknight that should resolve the bug and satisfy the ticket, although I'm sure checkdepend.t is going to say a curse word or two
19:03 whiteknight and then I'll pull out my checkdepend.t voodoo doll and jam a few more pins into it's groin
19:08 klavs left #parrot
19:08 NotFound I'm starting to think that embed.t and extend.t should be deleted. What they are doing is to block progress and "document" wrong ways of doing things.
19:09 klavs joined #parrot
19:20 whiteknight no argument here
19:20 whiteknight well, argument in so far as we want to test extending and embedding, but we want to test them correctly
19:22 soh_cah_toa joined #parrot
19:22 klavs left #parrot
19:23 soh_cah_toa ~~
19:30 fperrad left #parrot
19:30 lichtkind is gms and genms the same?
19:30 lichtkind soryy for delay was eating
19:30 whiteknight lichtkind: yes
19:30 lichtkind thanks
19:32 Kulag left #parrot
19:32 Kulag joined #parrot
19:35 soh_cah_toa whiteknight: take a look at the comment i left on TT #2134 you just opened
19:44 whiteknight soh_cah_toa: yeah, I saw that this morning. If you think you can clean that up, and you have the time to do so, please do
19:44 whiteknight I opened a few tickets last night, mostly as notes to myself as I was cleaning other stuff
19:45 whiteknight hardest part of that ticket is probably editing the makefile, then having to reconfigure/rebuild/re-run checkdepend.t to make sure everything is clean
19:45 whiteknight it will be instructive, at the least
19:45 soh_cah_toa ok
19:45 soh_cah_toa but does it actually use the embedding api? was i right or am i missing something?
19:46 whiteknight I don't know
19:46 whiteknight yeah, it uses the embedding API
19:47 whiteknight functions named "Parrot_api_*" are a dead giveaway
19:47 whiteknight the embedding API is the only thing that uses that naming scheme
19:47 soh_cah_toa ok good. i was confused there for a minute. that doesn't need to be done then. alright
19:47 whiteknight There's one API function, Parrot_api_disassemble_bytecode that we added only for this purpose. If we can find a way to refactor that logic out of libparrot entirely, that API can go away
19:47 whiteknight but I'll play with that later
19:47 soh_cah_toa ok sure
19:48 whiteknight What I think I would really like to do eventually is move the Parrot_disassemble function logic into a dynpmc somewhere, then rewrite pbc_disassemble in PIR
19:48 whiteknight less C code == more good
19:48 mikehh benabik: where did you add it - it really blows up for me
19:48 whiteknight but that's in the future
19:48 NotFound whiteknight: I'm thinking about adding a get_main method to the Eval PMC.
19:48 soh_cah_toa ok
19:49 whiteknight NotFound: that would be good. eventually I would like Eval PMC to go away
19:49 benabik mikehh: I just added `#include "parrot/parrot.h"` to the C template inside extend_vtable.t
19:49 NotFound whiteknight: yes, but we can use that same method in the appropiate packfile pmc.
19:49 whiteknight NotFound: right now, IMCC returns a PtrObj PMC for most cases, and that gets converted to an Evail in IMCCompiler PMC
19:49 mikehh benabik: at the beginning?
19:49 whiteknight NotFound: sure. That would be perfect
19:49 mikehh of the template that is
19:50 whiteknight Notfound: that's one of the last things we need to start writing the Parrot frontend in PIR. And I know it will help out some winxed issues
19:50 benabik mikehh: Yes, between <string.h> and "parrot/embed.h"
19:50 mikehh benabik: let me try that
19:52 whiteknight I ran that test before I pushed last night. I don't know why I didn't see those failures
19:52 NotFound I'm starting by cleening a beat eval.pmc and its pod.
19:52 NotFound bit
19:52 whiteknight NotFound: you can beat it too
19:53 bubaflub NotFound: i've got a couple Winxed questions if you've got a moment
19:53 NotFound bubaflub: sure
19:54 NotFound Just let me get some coffee ;)
19:54 bubaflub NotFound: of course.
19:54 benabik whiteknight: Might be differences in linkers.
19:54 dalek parrot: 668c468 | NotFound++ | src/pmc/eval.pmc:
19:54 dalek parrot: reorganize code, clean and improve pod in the Eval PMC
19:54 dalek parrot: review: https://github.com/parrot/parrot/commit/668c468cc8
19:56 NotFound Compiling with C++ assumes that undeclared funcions have C++ linkage, parrot public functions have C linkage, that is the problem.
19:58 NotFound bubaflub: ready
20:00 mj41 left #parrot
20:03 mikehh benabik: ok that worked for t/src/extend_vtable.t BUT NOT for t/src/embed.t and t/src/extend.t
20:04 bubaflub NotFound: great.  i've got a constructor that takes an optional var - i check if it's an Integer then call one initialize and set function, and if it's a string call a different initialize and set function.  That wasn't working so I swapped it out two separate functions - initialize the memory and then set it.  And still I'm not getting anything.  I'll gist something real quick for you to look at.
20:05 mikehh trying to add it to include/parrot/extend.h also blows up on all three tests (with g++)
20:05 mikehh but just those tests
20:06 benabik mikehh: Yes, I just started checking extend and embed.
20:07 bubaflub NotFound: https://gist.github.com/1023569 - i'm running basic.winxed and getting some unexpected output.  it seems like it's not really setting the integer.
20:07 mikehh whiteknight: the tests pass with gcc, just fails with g++
20:09 whiteknight loves me some g++
20:10 NotFound bubaflub: I think the problem is that your class is also called 'Integer'.
20:10 bubaflub NotFound: hmmm, ok.  I can name it GMPz or something like that.
20:10 mikehh adding parrot.h to extend_vtable.t works but fails with extend.t and embed.t
20:11 mikehh i.e directly adding to the c template code
20:12 dukeleto mikehh: whoa, why and how are you adding to parrot.h to those files?
20:13 NotFound bubaflub: uh, maybe not, I'm seeing that the output says "in here integer"
20:13 mikehh dukeleto: added at line 42 of t/src/extend_vtable.t
20:14 bubaflub NotFound: yeah, i believe it's detecting the var correctly.  for whatever reason it's not being set correctly.  but i can call mpz_set_ui (set a GMP integer from an unsigned int) in other places just fine.  and i can also initialize it just fine.  just for whatever the reason the combination is not happy.
20:15 NotFound bubaflub:  Maybe you do "self.ptr = ptr;" too late?
20:15 bubaflub NotFound: hmmm, maybe.  lemme try moving that real quick
20:15 bubaflub NotFound: do you think moving it to right after the _init function?
20:15 mikehh dukeleto: without it all tests fail with g++ build
20:16 NotFound bubaflub: yes
20:16 bubaflub NotFound: great, i'll try that.
20:16 bubaflub whiteknight: if you've got a moment, i'd like to ask you a few questions about setting up tests in Winxed.
20:17 mikehh dukeleto: Parrot_new not in scope
20:17 RobertLJ joined #parrot
20:18 bubaflub NotFound: nope, that didn't do it.
20:19 mikehh dukeleto: actually -> error: ‘Parrot_new’ was not declared in this scope
20:20 bubaflub NotFound: i think i got it fixed;  it's ugly but it works.
20:20 Coke M0?
20:20 mikehh and I am getting the same error with embed.t and extend.t
20:21 NotFound bubaflub: What's the problem?
20:21 mikehh adding parrot.h fixes the extend_vtable.t BUT not embed.t and extend.t
20:21 Coke aloha, M0?
20:21 aloha Coke: M0 is the VM that Lorito will run on top of. or http://leto.net/perl/2011/05/what-is-m0.html
20:21 Coke aloha, M0 glossy?
20:21 aloha Coke: Sorry, I don't know.
20:21 benabik aloha, lorito?
20:21 aloha benabik: lorito is "little parrot" in spanish or http://trac.parrot.org/parrot/wiki/Lorito or http://github.com/atrodo/lorito or http://github.com/ekiru/yalp-asm
20:23 bubaflub NotFound: apparently i was passing ptr instead of self
20:23 bubaflub NotFound: i'm not entirely sure why this works or even if it should; but somehow it *does* work
20:24 bubaflub NotFound: i'll update the gist with the latest that somehow works
20:25 whiteknight bubaflub: I am packing up and heading home soon. Can questions wait till later this evening?
20:25 bubaflub whiteknight: definitely.
20:26 whiteknight bubaflub: awesome. There are plenty of examples in the rosella t/ directory
20:26 bubaflub whiteknight: great, thanks.
20:26 bubaflub NotFound: https://gist.github.com/1023569
20:26 whiteknight most of the tests are written in NQP, but I added some example tests in winxed too: https://github.com/Whiteknight/Rosel​la/blob/master/t/winxed_test/Basic.t
20:27 whiteknight and the test harness is written in NQP, but it's so dirt simple that it's easy to translate into any language
20:27 bubaflub whiteknight: do i need to roll my own test harness?
20:27 whiteknight bubaflub: yes, but there's a Rosella harness library that does most of the work
20:27 whiteknight basically, you create a harness object, a testfile factory, and a view
20:28 bubaflub whiteknight: fantastic.  i'll take a look at it and hassle ya later tonight.
20:28 whiteknight then give it a list of files and/or a list of directories, and .run() it
20:28 whiteknight bubaflub: awesome. Definitely read through some of the stuff in the rosella t/ directory. It's all easier than you would expect
20:28 klavs joined #parrot
20:28 NotFound bubaflub: be careful of not mixing the nci functions that deal with the underlying gmp object and the ones that deal with the Winxed object.
20:28 * whiteknight is signing off now
20:28 whiteknight left #parrot
20:29 bubaflub NotFound: yeah, that's why i'm surprised this is working.  shouldn't i need to pass in ptr?
20:29 cotto_work aloha: m0 glossy is a brief self-contained glossy brochure-like introduction to M0 or https://gist.github.com/1022340 (draft)
20:29 aloha cotto_work: Okay.
20:29 bubaflub NotFound: though the Winxed object only wraps a single var ptr, perhaps that's why it is working.   but i'd rather do it right then have it accidently work
20:31 NotFound bubaflub: doesn't matter how simple it is, is a different object.
20:32 bubaflub NotFound: right.
20:33 bubaflub NotFound: so when i pass a GMP Integer var into a function, can i access the ptr directly?  e.g. x.ptr?  or do i need a method to return ptr?
20:34 hudnix joined #parrot
20:34 NotFound bubaflub: you can access the attribute directly.
20:34 bubaflub NotFound: ok.
20:43 dukeleto msg whiteknight can you do your refactoring of embed/extend stuff in a branch? You are moving code around which I am trying to get to certain test coverage goals to finish my grant
20:43 aloha OK. I'll deliver the message.
20:43 klavs left #parrot
20:49 lichtkind how many gc did you implement in last 2 years?
20:49 dukeleto lichtkind: a few ;)
20:51 lichtkind 3 maybe?
20:51 benabik Man // would be useful in NQP
20:52 dalek parrot-gmp: 3d262a7 | bubaflub++ | src/GMP. (4 files):
20:52 dalek parrot-gmp: fall back to use separate init and set funtions in the constructor
20:52 dalek parrot-gmp: review: https://github.com/bubaflub/​parrot-gmp/commit/3d262a71f4
20:52 dalek parrot-gmp: 6c745c5 | bubaflub++ | examples/basic.winxed:
20:52 dalek parrot-gmp: update example with constructing from integer and a string
20:52 dalek parrot-gmp: review: https://github.com/bubaflub/​parrot-gmp/commit/6c745c54a6
20:52 dukeleto lichtkind: we currently have 4, which you can find in src/gc/gc_*.c
20:52 dukeleto lichtkind: ms, ms2, inf and gms
20:52 lichtkind yeah but i cant believe you did more then 2 in last 2 years :)
20:53 dukeleto lichtkind: gms is the newest and default, and stands for "generational mark and sweep"
20:53 lichtkind thank you
20:53 dukeleto lichtkind: yes, ms2 and gms are the two recent gc's
20:53 cotto_work inf is mostly for debugging
20:53 dukeleto lichtkind: yes, inf is kind of like not having a gc. Useful for benchmarking and stuff
20:54 dalek parrot: a56af1b | NotFound++ | / (2 files):
20:54 dalek parrot: get_main method in Eval PMC to get the main sub, if any
20:54 dalek parrot: review: https://github.com/parrot/parrot/commit/a56af1b01e
20:54 dukeleto lichtkind: so we have 3 real GC algorithms in parrot core, currently
20:54 cotto_work I doubt ms has been tested at all recently
20:54 dukeleto cotto_work: fulltest exercises it, right?
20:55 cotto_work dukeleto: possibly, but nobody runs that except release managers
20:55 cotto_work and it doesn't run everything if one test fails
20:55 dukeleto cotto_work: suboptimal to say the least
20:56 dukeleto cotto_work: these are things we should be using the gcc compile farm for
20:56 cotto_work It feels bolted together rather than designed.
20:56 cotto_work dukeleto: +1
20:56 dukeleto cotto_work: we could easily have a machine running every commit through "fulltest"
20:56 dukeleto cotto_work: the problem is currently, where to do results go?
20:56 cotto_work We should also smarten up fulltest.
20:57 cotto_work To say "smolder" would be a punchline.
20:57 dukeleto cotto_work: how to you feel about a parrot-testers or parrot-smoke mailing list? Where it is encouraged to send test failure reports. Currently, smolder is down and people on parrot-dev don't want to see emails about failing tests
20:57 bubaflub hey dukeleto, wanted to let you know i just got string stuff working in my GSoC project as well as having a nice constructor.  i'll start the test suite tonight.
20:57 lichtkind what is the diff betweenmark and sweet 1 and 2
20:57 dukeleto bubaflub: wooo!
20:57 lichtkind what is the diff between mark and sweep 1 and 2
20:58 cotto_work dukeleto: sounds like a good idea
20:58 bubaflub dukeleto: yeah, even without the 't' NCI type it's not too bad.  i'll definitely write up some stuff on how to handle this
20:58 dukeleto lichtkind: a smarter version of the bacek++ coding robot wrote #2 :)
20:58 dukeleto bubaflub: great!
20:58 lichtkind bacek sounds czech
20:59 dukeleto lichtkind: he lives in Australia, methinks, but hails from the Russian lands
20:59 bubaflub he comes from the land of ice and snow
21:00 lichtkind thanks
21:00 dukeleto lichtkind: i am not sure of the differences between ms and ms2. If you send a quick email to parrot-dev, I am sure somebody will let you know :)
21:00 cotto_work diff src/gc/gc_ms.c src/gc/gc_ms2.c
21:00 cotto_work ;)
21:00 sorear lichtkind: he sometimes goes by Vasily Chekalkin (sp)
21:02 lichtkind are there any core sections other than gc largely rewritten or implemented since last 1,5h?
21:04 cotto_work lichtkind: since when?
21:04 dukeleto Coke: no, OSU is not on the smolder issue. It is up to us to fix smolder, and nobody has stepped up in many weeks. Doesn't bode well for smolder.
21:04 dukeleto lichtkind: since the last 1.h years?
21:04 lichtkind sorry i mean 1,5 years
21:04 sorear you mean, since 4.8 10^3 seconds ago?
21:05 lichtkind od course
21:05 sorear ah
21:08 lichtkind i heard something like all subsystems are there but in different qualities :)
21:09 lichtkind still the case?
21:09 cotto_work still the case
21:09 dukeleto lichtkind: the future is here, but it is unevenly distributed :)
21:11 lichtkind thank you for your support
21:16 lichtkind i also heard nothing from python and ruby compilers lately
21:19 dalek winxed: r1051 | NotFound++ | trunk/winxed_installed.winxed:
21:19 dalek winxed: Use get_main method in installable driver
21:19 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=1051
21:19 dalek winxed: r1052 | NotFound++ | trunk/pir/winxed_installed.pir:
21:19 dalek winxed: update generated installable driver
21:19 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=1052
21:23 ambs left #parrot
21:23 dalek parrot: 2ba966e | NotFound++ | ext/winxed/driver.pir:
21:23 dalek parrot: update winxed driver to use the get_main method in the Eval PMC
21:23 dalek parrot: review: https://github.com/parrot/parrot/commit/2ba966eb0b
21:24 contingencyplan joined #parrot
21:25 benabik lichtkind: lucian++ is working on Python on Parrot and rohit_nsit08 is working on JavaScript on Parrot.  I think someone was doing some hacking on Cardinal (Ruby on Parrot), but I don't know who.
21:29 lichtkind thank you
21:30 lichtkind sounded a bit that changing toolset and the thatswhy reimplementations discourages some people?
21:31 bubaflub lichtkind: Tene has been working on Cardinal, but he's going to use 6model (the Meta Object stuff from Rakudo) to do objects in Ruby
21:32 NotFound lichtkind: maybe, but people usually lose interest in his projects because of a variety of reasons or just without reasons, we can't know for sure.
21:33 Coke dukeleto: last I heard, OSU was getting it running again. if you need help getting it running, another email asking for help is probably a good idea.
21:34 Coke last time I tried to do something, OSU was in the middle of a big ... thing ... regarding smolder. I assumed the end result of that .... thing .... would be that smolder would be back up again.
21:34 Coke I'm happy to work on it as time permits.
21:34 estrabd joined #parrot
21:34 NotFound I see almost everyday lots of projects abandoned or unmaintained on the Internet without any explanation.
21:34 lichtkind NotFound: i know that i have a project since 7 years , its an ever coming and going
21:35 lichtkind even people that say they want to sjow up keep silent
21:35 NotFound I've seen pages with notes like this: "(C) 1998 - Under construction" X-)
21:35 lichtkind *haha*
21:36 lichtkind i just grabbing you insight since you closer to parrot core
21:36 mikehh NotFound: the trouble with Open Source is that people often move on, get bored or whatever, and bus numbers are far too low on many projects
21:37 NotFound Also with non Open.
21:37 cotto_work also with humans
21:38 mikehh it is particularly a problem with voluntary projects
21:38 NotFound Legolas and Gimli says something about that, but I don't have the quote at hand.
21:39 NotFound (even if I locate it, I don't have it in english)
21:40 estrabd left #parrot
21:41 PurityControl joined #parrot
21:43 lichtkind i have to admit parrot sounds much more interesting these days
21:45 davidfetter left #parrot
21:46 Patterner left #parrot
21:46 PurityControl is now known as Patterner
21:46 RobertLJ_ joined #parrot
21:47 bluescreen_ joined #parrot
21:47 bluescreen left #parrot
21:49 RobertLJ left #parrot
21:51 benabik I wonder why there's a FIXME in here saying "this line should go before this other line".  Why wasn't the line just moved?
21:51 fperrad joined #parrot
21:53 estrabd joined #parrot
21:55 estrabd left #parrot
21:55 Tene_ I haven't been actively working on cardinal; I keep meaning to, but never get around to it... :(
21:56 Tene_ some other people have been hacking on cardinal, though
21:57 estrabd joined #parrot
21:57 lucian_ joined #parrot
21:58 estrabd left #parrot
22:01 lucian left #parrot
22:02 bluescreen_ left #parrot
22:20 whiteknight joined #parrot
22:21 fperrad left #parrot
22:33 whiteknight NotFound: ping
22:34 kid51 joined #parrot
22:34 dalek parrot/whiteknight/embed_cleanups: 75aa901 | Whiteknight++ | src/ (2 files):
22:34 dalek parrot/whiteknight/embed_cleanups: More cleanups/removals in src/embed.c
22:34 dalek parrot/whiteknight/embed_cleanups: review: https://github.com/parrot/parrot/commit/75aa9016d5
22:37 lucian_ left #parrot
22:39 nopaste "kid51" at 192.168.1.3 pasted "All g++ build fails in master at a56af1b" (710 lines) at http://nopaste.snit.ch/51635
22:41 whiteknight I'm on it
22:41 whiteknight I need to reconfigure with g++
22:42 NotFound_b joined #parrot
22:42 NotFound_b whiteknight: pong
22:42 lucian joined #parrot
22:42 whiteknight NotFound: If we add a get_pointer vtable to Eval, we can use Parrot_pf_get_packfile_main_sub to get the main sub
22:43 whiteknight As much as possible, I would like to get packfiles more encapsulated
22:43 NotFound_b whiteknight: sounds good
22:44 NotFound_b And reimplement get_main using it, to avoid duplicate code?
22:46 whiteknight that's my thought
22:46 whiteknight have you added tests for get_main?
22:47 NotFound_b But Eval inherits from Sub, and Sub used get_pointer
22:47 whiteknight oh
22:47 whiteknight crap
22:47 NotFound_b We must check if overriding it causes some problem.
22:47 whiteknight we won't be able to do eval()
22:47 NotFound_b Uh
22:48 whiteknight actually, I don't know what Sub.get_pointer does
22:48 NotFound_b return sub->seg->base.data + sub->start_offs;
22:50 NotFound_b invoke does not use it.
22:50 whiteknight I don't like how Eval.get_string returns the serialized bytecode of the eval
22:51 whiteknight that's....ugly
22:51 whiteknight very ugly
22:51 whiteknight We should be using a PackFile PMC if we want to be serializing packfiles
22:51 NotFound_b Yes, but if we plan to deprecate Eval there is no point in trying to improve it now.
22:52 jnthn__ ...deprecate...Eval?
22:52 jnthn__ Not until you have a replacement you don't.
22:52 jnthn__ A replacement that meets my needs.
22:53 NotFound_b jnthn__: Eval does too much things, we need something more controllable.
22:53 jnthn__ Being able to iterate through the compiled Subs, like Eval can today, is included in those.
22:53 jnthn__ NotFound_b: Fine, but please deprecate it *after* introducing that controllable thing and making sure it meets folks needs.
22:53 NotFound_b jnthn__: we should be able to do that with a packfile PMC.
22:53 whiteknight jnthn__: yes, those capabilities are included in bigger packfile refactors anyway
22:53 whiteknight er, exposing those capabilities
22:54 NotFound_b jnthn__: yes, that is the plan, doing it easy.
22:54 jnthn__ OK, good. :)
22:54 whiteknight TT #1969
22:54 whiteknight no movement on it yet, it isn't in api.yaml, but we are working towards it
22:56 whiteknight IMCCompiler has .compile_string and .compile_file methods, those are experimental and won't be returning Eval for long
22:56 whiteknight compile_file already doesn't return an Eval, and compile_string is mostly unused
22:57 whiteknight The PackFile PMC currently is used for building packfiles, not introspecting existing packfiles. We need to add that capability
22:57 whiteknight once we have that, we should be good to start a transition
22:58 jnthn__ OK. I just get nervy about "we're deprecating X" when there isn't a replacement that I can migrate to.
22:58 jnthn__ s/when/before/
23:03 NotFound_b jnthn__: soemtimes we confuse "deprecate" with "remove a deprecated thing"
23:04 NotFound_b "Deprecate" should be document the fact, how to avoid using the deprecated thing, and maybe emiting a warning when used,
23:05 sorear if you deprecation-notice something without giving a replacement, what good does it do?
23:06 NotFound_b sorear: sometimes there is no replacement, is just "Don't do that".
23:07 jnthn__ NotFound_b: Yes, but if "that" is something a Parrot user is relying on, that's an outstanding way to lose a userbase.
23:07 NotFound_b Ideally, very few times.
23:07 mtk left #parrot
23:08 kid51 NotFound_b: What email address is best to use for comments/discussion about winxed and/or its website?
23:08 NotFound_b kid51: http://winxed.org
23:08 NotFound_b kid51: I added today an email address to the main page.
23:08 kid51 Thx
23:09 sorear NotFound_b: "that" in this context is runtime compilation of PIR
23:09 jnthn__ sorear: I'm not sure NotFound_b was implying that was the reality for this csae. Or at least, I hope not. :)
23:09 NotFound_b sorear: We have deprecated that?
23:09 NotFound_b Oh, I see.
23:11 NotFound_b No, the idea is that compilers will give other thing instead of an Eval PMC. A packfile PMC, probably.
23:12 NotFound_b And provide in that PMC all interfaces needed.
23:13 NotFound_b But maybe running it as a Sub is not desirable.
23:16 lichtkind i overlooked rosella
23:16 lichtkind since when exist that?
23:19 sorear last month
23:19 NotFound_b whiteknight: Parrot_pf_get_packfile_main_sub takes a packfile PMC, not a PackFile struct. We don't have it in a Eval.
23:21 dukeleto ~~
23:22 lichtkind sorear: and got it used so quickly?
23:22 dukeleto lichtkind: rosella has been around for at least a few months now
23:22 lichtkind is it some technology to help you through the loreto rewrite?
23:23 sorear no
23:23 sorear rosella is completely unrelated to lorito
23:23 benabik Other than both of them being awesome.
23:24 NotFound_b lichtkind: is a bunch of general purpose libraries packed together.
23:25 lichtkind NotFound_b: yeah i read that on the page but sounds like its really used by compilers or isnt it?
23:26 NotFound_b I think lucian is using or planning to use it for the tests of its python implementation.
23:26 lucian yes, i do use rosella/test for now
23:27 lucian which reminds me, i should get off my ass and write a harness for it
23:28 lucian i also used NotFound_b's/whiteknight's prototype object .. uh.. prototype in rosella/unstable/prototype
23:28 lucian as an example
23:28 dmalcolm left #parrot
23:30 dukeleto lichtkind: http://whiteknight.github.com/Rosella/
23:32 lichtkind thanks
23:32 dukeleto lichtkind: i would describe it as a toolkit for Parrot HLL's that provides basic tools like test suites, mock objects, memoization and such
23:33 dukeleto lichtkind: so it sits above Parrot core and makes parrot easier to use for HLL developers. It is an "easy onramp"
23:34 kid51 is now known as kid51_at_dinner
23:39 lichtkind dukeleto: was this not the task for PCT ?
23:43 bubaflub lucian: you're using rosella tests?  i think i'm going to use that as well to test my Winxed code
23:44 dukeleto lichtkind: PCT are compiler tools, for actually writing a HLL. Rosella are tools that HLLs use, for instance, to write or run tests
23:44 dukeleto lichtkind: so Rosella lives on top of HLLs and PCT is below HLLs (i.e HLLs are built on top of PCT)
23:45 dukeleto bubaflub: rosella's test stuff looks pretty nice
23:45 NotFound_b lichtkind: the language creation scripts creates a skeleton, but they are just helpers, not an important part of PCT.
23:46 bubaflub dukeleto: agreed. i'm a little hesitant about having to roll a lot of my own stuff, but hopefully whiteknight is around to help me out.
23:46 bubaflub speaking of, ping whiteknight
23:47 dukeleto bubaflub: what kind of own-stuff rolling?
23:48 bubaflub dukeleto: it looks like i need to make my own harness and a few other things
23:48 dukeleto bubaflub: really?
23:48 dukeleto bubaflub: http://whiteknight.github.com/​Rosella/libraries/harness.html
23:48 dukeleto bubaflub: you using that?
23:49 bubaflub dukeleto: http://irclog.perlgeek.de/p​arrot/2011-06-13#i_3918803
23:49 bubaflub dukeleto: yeah, i guess i just need clarification on exactly what i need from Rosella.  i'm use to a Test::More type dealio
23:51 dukeleto bubaflub: https://github.com/Whiteknight/parrot​-linear-algebra/blob/master/t/harness
23:51 dukeleto bubaflub: looks like you can copy that and tweak
23:51 bubaflub dukeleto: cool.  i'd rather not do it in NQP and just stick with Winxed but yes, that looks good.
23:52 dukeleto bubaflub: sure, you can use it from Winxed
23:52 dukeleto bubaflub: there may be a winxed harness using rosella sitting around
23:52 bubaflub dukeleto: great.  i'll get cracking on this.
23:53 dukeleto bubaflub: but Rosella (written in NQP) chooses to use NQP for the harness: https://github.com/Whiteknight​/Rosella/blob/master/t/harness
23:53 dukeleto bubaflub: so maybe don't worry too much about the language of your test harness
23:53 dukeleto bubaflub: having a working harness is more important
23:53 bubaflub dukeleto: ah, ok.  i suppose as long as it runs with an installed parrot, i'm happy.
23:53 NotFound_b Rosella is language agnostic AFAIK
23:54 dukeleto bubaflub: yep. nqp is in parrot core.
23:54 bubaflub dukeleto: fantastic.  i need to move a few bits and pieces around, but i should have some tests running tonight.
23:54 dukeleto bubaflub: good to hear
23:54 whiteknight bubaflub: pong
23:55 bubaflub ah, hola whiteknight.  i was just going to ask you for more info about Rosella testing stuff.
23:55 dukeleto whiteknight: i've been talking up your Rosella library. It is really looking shiny these days.
23:55 whiteknight dukeleto: thanks! I need all the PR I can get
23:55 dukeleto whiteknight: the website is slick
23:56 bubaflub whiteknight: i've got NCI bindings and a Winxed class that wraps around it; i think Rosella is the way to go but just need some guidance.  from what i understand, i need a Test Harness that tells it what directories/files to load, individual test files, and then just dump out some results, right?
23:56 whiteknight dukeleto: thanks!
23:56 whiteknight bubaflub: yes. Test harness runs the tests. It's dirt-easy to write
23:56 bubaflub whiteknight: cool.  i'm looking at your parrot-linear-algebra one in NQP.
23:57 bubaflub whiteknight: do i need a checkout of Rosella in my repository?
23:57 whiteknight bubaflub: no. You can just install it. You have plumage? "plumage install rosella"
23:57 bubaflub whiteknight: yes, i have plumage.  i'll run it now
23:58 bubaflub hmmm, running `plumage` gives me: PARROT VM: Could not initialize new interpreterunimplemented value type -98
23:58 bubaflub maybe i need to re-install it
23:58 whiteknight bubaflub: http://whiteknight.github.com/2011/0​3/30/rosella_harness_refactors.html
23:59 whiteknight bubaflub: yes, probably need to uninstall/clean/rebuild/reinstall plumage
23:59 bubaflub whiteknight: yep, that was it.  must have installed it a long time ago.

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

Parrot | source cross referenced