Camelia, the Perl 6 bug

IRC log for #parrot, 2009-08-14

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 cotto the destructor never gets called
00:01 bacek joined #parrot
00:01 chromatic Ah, hang on a moment.
00:03 dalek parrot: r40535 | chromatic++ | branches/pluggable_runcore (2 files):
00:03 dalek parrot: Fixed enabling of the profiling core; initialization code goes in the ->runops
00:03 dalek parrot: pointer, confusingly -- preparing for a run is only effective for certain types
00:03 dalek parrot: of cores which need special bytecode setup.
00:03 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40535/
00:04 chromatic Now for destroy.
00:04 chromatic There's the problem.  Parrot_runcore_destroy() only gets called if there are dynop libraries loaded.
00:05 patspam joined #parrot
00:06 cotto That explains what you meant by "but don't forget to tailcall that function"
00:07 * cotto didn't read carefully enough
00:07 chromatic I forgot to do that too.
00:11 chromatic I should have the tailcall here in a moment.
00:13 dalek parrot: r40536 | chromatic++ | branches/pluggable_runcore/src (2 files):
00:13 dalek parrot: Enable runcore destruction only in parentmost interpreter; ensure destruction
00:13 dalek parrot: regardness of presence of Parrot's really destroy CLI flag.
00:13 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40536/
00:15 chromatic That should do it.
00:16 dalek parrot: r40537 | chromatic++ | branches/pluggable_runcore (2 files):
00:16 dalek parrot: [pluggable_runcore] Fixed up profiling core first-time initialization,
00:16 dalek parrot: including a call to the actual "Run profiling ops!" step.
00:16 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40537/
00:23 cotto It's a little too eager now.  The profiling destructor gets called when I'm not using the profiling runcore.
00:26 chromatic Ah, then you know the pointer dance to do and where.
00:29 cotto Oh.  It does that explicitly.  Should I just make the destructor smart enough to do nothing when the profiling runcore hasn't been initialized?
00:30 cotto nm.  Not what I'd call intuitive, but it works.
00:33 chromatic I'd leave the destroy pointer empty until you've initialized the runcore.
00:33 dalek parrot: r40538 | cotto++ | branches/pluggable_runcore/src/runcore/cores.c:
00:33 dalek parrot: [profiling] do a special pointer dance so the profiling runcore's destructor is only called when the profiling runcore has been initialized
00:33 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40538/
00:33 chromatic Just like that.
00:34 cotto do you mind if I update from trunk?
00:35 chromatic Not at all.
00:35 Whiteknight are Runcores a PMC type?
00:36 Whiteknight I didn't think I heard that, but all this talk of destructors is making me wonder
00:36 Whiteknight and if they are not currently PMCs, that's my idea and you can't have it
00:37 cotto They're not.
00:37 chromatic I don't think they should be.
00:39 dalek parrot: r40539 | cotto++ | branches/pluggable_runcore/tools/dev/pprof2cg.pl:
00:39 dalek parrot: [profiling] various updates to the profile postprocessing script (it's still broken, though)
00:39 quek left #parrot
00:39 chromatic They don't really support PMC-esque operations, for one.  They'll be a bear to convert to Lorito if they're PMCs for two.
00:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40539/
00:39 dalek parrot: r40540 | cotto++ | branches/pluggable_runcore/src/runcore/cores.c:
00:39 dalek parrot: [codingstd] fix a codingstd nit
00:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40540/
00:41 Whiteknight but there are plenty of object types that don't really fit into the VTABLE interface
00:41 Whiteknight at least not well
00:41 cotto Speaking of Lorito, I'm concerned about it ever getting past the "iwbn" stage.
00:41 Whiteknight All you would really need for a RunCore PMC would be init, destroy, and invoke
00:41 Whiteknight iwbn?
00:41 Whiteknight oh, it would be nice
00:42 chromatic Turns out you need more than that for a RunCore PMC.
00:42 cotto The LoritoPrimitives page hasn't attracted any Smalltalkers and I'm not sure what else would be a good way to define a set of what ops Lorito will need.
00:42 Whiteknight oh, so I'm probably radically oversimplifying
00:43 Whiteknight cotto: I think the best way to go about it would be to look through C and determine what kinds of operational primitives that language uses
00:44 chromatic I have my Smalltalk books right here.  I just haven't had the spare cycles.
00:45 Whiteknight basic arithmetic primitives, pointer reference and dereference, pointed-based indexing, pushing and popping arguments, calling a function. These are probably the most important
00:46 chromatic Basically.
00:47 Whiteknight anything else, while not ideal, we could do inside a function call
00:47 chromatic Or in terms of the other primitives.
00:47 Whiteknight right
00:48 Whiteknight also type conversions, especially INTVAL <-> FLOATVAL
00:48 Whiteknight and if we can't resolve the dependencies, INTVAL <-> PT
00:48 Whiteknight PTR*
00:49 cotto merging?
00:49 cotto merging is slow
00:49 cotto merging?
00:49 purl i think merging is slow
00:50 GeJ Brace! Brace!"
00:57 tetragon joined #parrot
00:59 bacek_at_work good morning
00:59 purl Here I am, brain the size of a planet, and all they say is 'Good Morning'
00:59 bacek_at_work purl: bad girl
00:59 purl bacek_at_work: sorry...
00:59 GeJ G'Day bacek
01:01 cotto hi bacek_at_work
01:08 bacek_at_work G'Day GeJ
01:08 bacek_at_work G'Night cotto
01:08 cotto It's not that late.
01:08 Whiteknight good morning bacek
01:11 bacek_at_work Mornin' Whiteknight
01:13 dalek TT #913 created by basic++: Test123
01:18 cotto I don't think that's a real ticket.
01:25 darbelo It claims to be "Just another email2trac test", maybe that means email2trac works now.
01:25 cotto It's nice to have that merge done.
01:26 cotto now I get to experience the shininess of the corevm make target
01:37 dalek parrot: r40541 | cotto++ | branches/pluggable_runcore (459 files):
01:37 dalek parrot: bring branch up-to-date with trunk
01:37 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40541/
01:37 chromatic Maybe I should have tried to use git for that merge.
01:37 ilbot2 joined #parrot
01:37 Topic for #parrotis now HAPPY BIRTHDAY #PARROT! | http://www.parrot.org/ | 1.4.0 | For 1.5: Remove Deprecated Features | http://planet.parrotcode.org/
01:37 japhb kthakore: Better than not working ....
01:37 kthakore thx
01:39 kthakore well considering I started learning perl 3 months ago I think I am doing ok ... ish
01:39 japhb Three months and already fixing SDL?  I'd say that's pretty decent.  :-)
01:40 kthakore japhb, thats how I learn hacking
01:40 nperez joined #parrot
01:40 japhb And a good way it is.
01:40 japhb Collateral benefits.
01:41 kthakore that too
01:41 kthakore and having people look at my code, makes me want to put out better stuff
01:42 kthakore so its not like a school project that I do half ass
01:44 kid51 joined #parrot
01:45 GeJ cotto++
01:46 japhb kthakore: Misspent youth: http://www.perl.com/lpt/a/945
01:46 japhb Too bad it took WAY too much time to write those articles.
01:46 preflex joined #parrot
01:47 kthakore japhb, ooh that reminds me who worked on the events binding for SDL code
01:47 kthakore japhb, that stuff needs overhaul huge
01:47 japhb Man, redoing all that benchmarking with NYTProf would be a fun exercise.  ;-)
01:47 kthakore NYTProf ?
01:47 purl NYTProf is a *profiler*
01:48 kthakore kk
01:48 japhb Devel::NYTProf is the current state of the art in Perl 5 profiling -- just leaps and bounds ahead of DProf.
01:49 chromatic DProf segfaults too often for anyone to have ever considered it state of the art.
01:49 kthakore I have yet to run bindings that have perl overhead through a profiler
01:49 kthakore But optimization comes last
01:49 kthakore as they say
01:50 kthakore chromatic, don't you mean NYTProf
01:50 japhb chromatic: No argument.  But at the time, it met the need.
01:50 * kthakore looks at perldoc 0_o
01:50 dalek TT #914 created by basic++: email2trac test
01:50 japhb kthakore: He meant, no one should take my statement to mean that DProf was the state of the art at any point before the arrival of NYTProf.  :-)
01:51 kthakore oh ok
01:51 kid51 What does the NYT in that module's name stand for?
01:51 kthakore anywho optimization will come later atleast for me
01:51 * kid51 always thinks New York Times or New York Tech ...
01:53 japhb kid51 is right.  New York Times sponsored the initial work on it.  (Well, sponsored in the sense of 'the guy who worked on it was working for them'.)
01:54 dalek TT #913 closed by jkeenan++: Test123
01:54 dalek TT #914 closed by jkeenan++: email2trac test
01:54 kid51 Really?  I was just fantasizing that because I type those 3 letters into my browser so often.
01:54 kid51 Never dreamt it was the real explanation.
01:54 japhb :-)
01:55 japhb Best. Advertising.  Ever.
01:55 kid51 Sad to say, but the module may outlive the paper.
01:57 dalek TT #915 created by basic++: email2trac test
01:57 dalek parrot: r40542 | cotto++ | branches/pluggable_runcore (2 files):
01:57 dalek parrot: [profiling] start putting runcore-specific data into a Parrot_profiling_runcore_t and using init/destroy
01:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40542/
01:57 * japhb decommute
02:00 dalek TT #914 reopened by basic++: email2trac test
02:02 kid51 Here's a Smolder report on the pluggable_runcore branch:  http://smolder.plusthree.com/app/pu​blic_projects/report_details/26135
02:02 kid51 done at r40537
02:05 chromatic Heh.
02:05 chromatic One fprintf removal could fix that.
02:05 kthakore gdnite
02:11 nathanmccauley joined #parrot
02:11 basic joined #parrot
02:12 basic Sorry for the spam, I got excited it finally worked
02:13 basic kid51: I've been battling the email2trac script for the parrot trac instance for what seems like a year, it is working now, but to test I had to send some emails
02:13 kid51 That's great.  Perhaps a bit more content in them explaining what you were up to would have helped.
02:14 kid51 Are you with Trac?
02:14 cotto Meh.  I'm just glad that it's working.  Nice work basic!
02:14 basic Sorry about that, I sent an email to the ticket we had, but only 3 of you were on it
02:14 basic I didnt realize it was going to spam the channel
02:14 basic cotto: thanks! sorry it took so long
02:15 basic kid51: i'm with OSUOSL, we host the VM and I'm the Trac guy here
02:16 kid51 Ah, good!  Am looking at osuosl.org
02:16 basic :)
02:16 basic I wasn't really expecting the emails to actually go through for once
02:19 kid51 Hmm, perhaps Parrot should be listed as one of the projects you host:  http://osuosl.org/services/hosting/communities
02:19 basic It's not on there? I'll bug our operations manager about that
02:20 basic It's a pretty big one :)
02:20 kid51 Well, there's nothing between "Oregon State Linux Users group" and "Participatory Culture foundation"
02:21 basic kid51: looks like it's not there, i just let Jeff Sheltren know (our operations guy)
02:21 kid51 Of course, if it's our fault for not supplying you with a logo and copy, let us know.
02:21 japhb joined #parrot
02:22 basic sure, i'll get back to you on that
02:22 kid51 Logistical question for any of our admins:  If I wanted to start a mailing list called 'nycparrotusers', could that be set up to be 'nycparrotusers@parrot.org'?
02:23 kid51 ... i.e., are we prepared to host Parrot-related lists that aren't quite the Parrot project itself?
02:41 janus joined #parrot
02:45 * kid51 must sleep
02:45 purl $kid51->sleep(8 * 3600);
02:47 nopaste "GeJ" at 202.171.79.162 pasted "[pluggable_runcore] Patch MANIFEST file to reflect the new pprof2cg.pl tool" (41 lines) at http://nopaste.snit.ch/17547
02:48 nathanmccauley_ joined #parrot
02:50 GeJ this patch fixes some failure in t/manifest/02-regenerate_file.t
03:20 donaldh joined #parrot
03:54 quek joined #parrot
04:34 eternaleye joined #parrot
04:50 TiMBuS joined #parrot
05:23 szabgab_ joined #parrot
05:36 quek left #parrot
06:02 tetragon joined #parrot
06:16 ash_ joined #parrot
06:16 uniejo_ joined #parrot
06:21 HG` joined #parrot
06:25 theory joined #parrot
07:11 tetragon joined #parrot
07:15 nathanmccauley joined #parrot
07:20 donaldh joined #parrot
07:28 cotto GeJ, thanks for that.
07:30 GeJ cotto: you're welcome.
07:30 GeJ already or still awake?
07:30 GeJ clock?
07:30 purl GeJ: LAX: Fri 12:30am PDT / CHI: Fri 2:30am CDT / NYC: Fri 3:30am EDT / LON: Fri 8:30am BST / BER: Fri 9:30am CEST / IND: Fri 1:00pm IST / TOK: Fri 4:30pm JST / SYD: Fri 5:30pm EST /
07:31 dalek parrot: r40543 | cotto++ | branches/pluggable_runcore/MANIFEST:
07:31 GeJ that must be "still" then.
07:31 dalek parrot: [MANIFEST] add tools/dev/pprof2cg.pl to the manifest
07:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40543/
07:33 GeJ cotto: codingstd tests barf in several places (mostly due to pprof2cg.pl and some .c files that were heavily modified today).
07:33 GeJ Is it ok if I prepare a couple of nopastes and msg them to you?
07:34 cotto no need to msg them.  Also, pprof2cg.pl is *very* alpha.  I'm not sure it it's worth fixing at this point.
07:34 cotto and by "alpha", I mean broken
07:35 cotto and the profiling runcore seems to be exploding rather spectacularly atm
07:36 GeJ I don't pretend I can fix the logic or the features... I was just suggesting to lint it so I can hush `make fulltest`.
07:36 chromatic Is Saturday ambitious for a trunk merge?
07:37 chromatic If you remove the fprintf from the destroy function, you'll pass more tests....
07:38 cotto What needs to happen for a merge?  I can get the code stable by then, but getting a good profile of Rakudo hello world that Callgrind can use is further off.
07:39 chromatic How good is "good"?
07:39 MinorToken joined #parrot
07:40 cotto sensical
07:40 chromatic What's left to do there?
07:40 cotto i.e. no percentages above 100 and no %NAN.
07:41 chromatic Is that the Perl 5 program or the reporting?
07:41 cotto in what Callgrind displays
07:41 chromatic Where do you think the error is?
07:41 cotto inner runcores are certainly a large part of the problem
07:42 chromatic Ah, especially from IMCC.
07:43 cotto The fprintf has been removed from the destroy function.
07:43 chromatic Fortunately we should stay in the profiling core through runloop switches.
07:44 cotto d'oh.  Apparently allocating a Parrot_runcore_t and casting it to a Parrot_profiling_runcore_t isn't good enough.
07:44 cotto Yes, it seems we do.
07:45 cotto That fixes the spectacular explosion.
07:46 cotto Do you want the profiling runcore to be usable before the merge or should it just not mess anything else up?
07:46 chromatic I'd like to be reasonably usable.
07:46 cotto ok.  I'll see how far I can get and try to think of ways the work can be parallelized.
07:46 chromatic The code cleanup is nice, but it's an internals only thing.
07:47 cotto yeah
07:47 chromatic If we just can't coax any useful or reliable information out of the runcore before then, extra time to see how things work with other languages won't hurt.
07:47 chromatic ... but despite the size of the changes, I don't consider them risky.
07:47 chromatic If I didn't get things right, *nothing* would work.
07:52 eternaleye chromatic++ # pluggable runcores
07:53 chromatic Not quite pluggable yet; there's no dynruncore support at the moment.
07:54 chromatic Let's call them "Mostly Encapsulated" for now; I'm an editor.  I can cut down the full description to two words.
07:54 chromatic That branch'll get destroyed for an interstellar bypass any day now anyway.
07:56 * cotto looks at code and reminds himself "clarity is better than cleverness"
07:58 einstein joined #parrot
07:59 tetragon joined #parrot
08:03 iblechbot joined #parrot
08:56 sjn joined #parrot
09:19 muixirt joined #parrot
09:20 PacoLinux joined #parrot
09:31 PacoLinux joined #parrot
09:36 whoppix joined #parrot
09:41 tetragon joined #parrot
09:52 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r40543 - Ubuntu 9.04 amd64
10:09 tetragon joined #parrot
10:10 mikehh rakudo (0d4fe08) builds on parrot r40543 - make test/make spectest (up to 27988) PASS - Ubuntu 9.04 amd64
10:10 eternaleye joined #parrot
10:20 athomason joined #parrot
10:27 japhb joined #parrot
10:47 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r40543 - Ubuntu 9.04 amd64 (built with gcc - prev test with g++)
10:51 kid51 joined #parrot
10:53 szabgab_ joined #parrot
11:12 tetragon joined #parrot
11:15 quek joined #parrot
11:20 donaldh joined #parrot
11:29 donaldh left #parrot
11:40 masak joined #parrot
11:41 AndyA joined #parrot
11:49 rdice joined #parrot
11:50 payload joined #parrot
12:01 tetragon joined #parrot
12:19 tetragon joined #parrot
12:20 ruoso joined #parrot
12:52 bkuhn joined #parrot
13:18 Coke joined #parrot
13:18 Coke AIGH
13:18 Coke oh, wait, I'm supposed to think that part in silent frustration.
13:18 Coke (not parrot related)
13:20 Infinoid on IRC, it's hard to disambiguate that from other forms of silence :)
13:21 particle don't we have smolder to report those results posted here overnight?
13:26 donaldh joined #parrot
13:40 * Coke wonders, shouldn't his consulting company know exactly how many hours of PTO he has remaining? Why ask him?
13:41 dukeleto Coke: that means you take how many hours you actually have left and multiply by 4 ;)
13:50 dukeleto mornin'
14:02 JimmyZ joined #parrot
14:10 dalek TT #531 closed by dukeleto++: [GSOC] Decide on a Pilot/Co-Pilot for the Parrot department of TPF this ...
14:13 Coke dukeleto: if only I knew how many I started with. cluster****
14:14 * Coke smacks perljam.
14:16 PerlJam whatdidido?
14:17 PerlJam or should I interpret that "smack" as a kiss and just blush and say thanks?  :)
14:18 PacoLinux joined #parrot
14:20 dalek TT #722 closed by dukeleto++: Rewrite of t/pmc/capture.t in PIR
14:22 Coke PerlJam: are you familiar with Duff's Device ?
14:22 PerlJam yeah, but it's not my fault :)
14:22 Coke ok. now look at your last facebook status.
14:23 PerlJam oh, heh
14:23 * Coke smacks perljam.
14:24 Coke Sorry. dog must know WHY he's being smacked or it won't work. =-)
14:25 * PerlJam makes a mental note to choose which Eagles lyrics he posts on FB more carefully.
14:26 Coke RAR.
14:26 dalek parrot: r40544 | dukeleto++ | trunk (2 files):
14:26 dalek parrot: [TT #871] Add srand() as a dynop
14:27 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40544/
14:27 dalek TT #916 created by dukeleto++: Need something like PMCNULL for STRINGs
14:30 quek left #parrot
14:44 Coke dukeleto++ #busy guy
14:45 Coke mikehh: with my working repository with a bandaid over the "getListValue()" bug, I seem to be running most of the spec test. will let it finish, check in, then you can add that to your rotation if youlike. =-)
14:47 dalek parrot: r40545 | NotFound++ | trunk/src/pmc/array.pmc:
14:47 dalek parrot: [cage] drop empty Array.class_init
14:47 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40545/
14:48 Psyche^ joined #parrot
14:52 dukeleto Coke: i slept 12 hours last night and had a burst of activity this morning
14:53 Coke if you ever get really bored, feel free to fix all the segfaults in "make spectest" on partcl. =-)
14:53 particle dukeleto: i don't think i've ever seen you sleep more than five hours
14:54 * PerlJam wonders why particle watches dukeleto when he sleeps
14:54 particle we're former conference roommates
14:54 dukeleto particle: that is because I stayed up until 3am 2 nights ago add the rand() dynops, so I went to sleep directly after work (without eating dinner) at 6:30pm and woke up this morning at 6:30am. I call it rebooting my biorhythm.
14:54 dukeleto particle watches everyone as they sleep, doncha know
14:55 particle i see you when you're sleeping, i know when you're awake
14:55 PerlJam particle: funny ...  I don't recall you looking like a fat man in a red suit.
14:56 particle i am large. i contain multitudes.
14:58 Coke dukeleto: biorrhyea.
15:00 Coke (fat man in red suit.) that's me, Diado Coleda.
15:05 jan joined #parrot
15:07 particle happy birthday, diado!
15:07 Coke does svn support... symlinks?
15:07 dukeleto Coke: probably not in a portable way
15:07 Infinoid http://subversion.tigris.org/faq.html#symlinks
15:07 Coke partcl's current setup lists all the skips/reasons in the wiki, and then the source repo uses that to figure out what spec tests to run.
15:08 Coke I was thinking about avoiding the separation there and putting the skip list into the source.
15:08 Coke guess I can do that and just have a wiki page that links to the skip list.
15:08 Coke (then no svn linkage)
15:09 eirik joined #parrot
15:16 Coke hey, a completely different set of tests are segfaulting. whee. =-)
15:16 Coke (actually, (#*@$&@#)
15:16 quek joined #parrot
15:17 mikehh Coke: I will test again as soon as you commit
15:19 dalek TT #917 created by jkeenan++: What mailing lists should parrot.org host?
15:20 donaldh joined #parrot
15:23 Coke mikehh: I need to finish the test, then update the SKIP list, then commit. (then I can rerun to get an "official" run.) PITA.
15:31 NotFound Coke: partcl's PMC class_init have a if (pass) that's no longer needed.
15:34 Coke Thanks.
15:41 PacoLinux joined #parrot
15:50 JimmyZ joined #parrot
15:59 iblechbot joined #parrot
16:01 ash_ joined #parrot
16:04 NotFound Coke: I was doing some checks, planning to change the handling of the init_class to avoid have in scope unrelated variables like that.
16:14 PacoLinux joined #parrot
16:21 riffraff joined #parrot
16:22 PacoLinux joined #parrot
16:31 bacek joined #parrot
16:41 payload joined #parrot
16:43 mokurai joined #parrot
16:58 PacoLinux joined #parrot
17:13 dalek parrot: r40546 | NotFound++ | trunk/src/pmc (3 files):
17:13 dalek parrot: [cage] drop no longer useful "if (pass)" conditions from class_init functions
17:13 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40546/
17:21 dalek rakudo: 58bef37 | jnthn++ | src/pmc/p6invocation.pmc:
17:21 dalek rakudo: Detect a Null PMC Access condition that could occur and throw a more meaningful exception.
17:21 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/5​8bef37b66ce0ba36553584e92e166dd3d7ed5df
17:21 dalek rakudo: 4a22048 | jnthn++ | src/ (2 files):
17:21 dalek rakudo: Refactor method dispatch code generation so that .*@foo and other similar things are no longer an error, which passes us some more tests. Also refactor .?, .+ and .* to work in terms of primitives provided by existing dispatchers, so we can rip out a dispatch helper, which duplicated the functionality they provided.
17:21 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​a220483fa6f7717e2e0a9f7def72e7c242161c8
17:24 chromatic joined #parrot
17:29 basic Coke: kid51 noticed Parrot wasn't on our communities page last night, we just got it added with the description from parrot.org. -- https://osuosl.org/services/hosting/communities
17:30 basic Jeff Sheltren says he's sorry for not adding it sooner
17:30 Coke basic;danke
17:30 basic Coke: you're welcome, let me know if you need anything changed around, sorry again for the confusion last night - i wasn't expecting it to work :)
17:35 Coke I wasn't confused. =-)
17:36 Coke but yah, there was some. NP.
17:44 dalek parrot: r40547 | NotFound++ | trunk/src/dynpmc/gdbmhash.pmc:
17:44 dalek parrot: [cage] drop no longer useful "if (pass)" condition from gdbmhash class_init
17:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40547/
17:54 gigabo joined #parrot
18:01 PacoLinux joined #parrot
18:03 dalek TT #918 created by NotFound++: Change handling of init_class functions
18:18 dalek partcl: r582 | coke++ | wiki/SpecTestStatus.wiki:
18:18 dalek partcl: update SKIP list
18:18 dalek partcl: (these segfaults are killing us.)
18:18 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=582
18:24 quek left #parrot
18:41 chromatic NotFound, re #918, all of that code using the init variable is generated code.  Anyone writing PMCs by hand is crazy.
18:45 NotFound chromatic: the seudo-vtable class_init is from the .pmc file
18:46 NotFound I mean init_class
18:47 NotFound Ugh, I always confuse me with that names.
18:48 NotFound I'll fix the ticket in a moment.
18:48 chromatic It is?  I thought we generated all of those.
18:48 chromatic ... except for singleton PMCs.
18:49 NotFound The body of the function in the .pmc is inserted into the generated one.
18:49 ash_ joined #parrot
18:49 NotFound Array has one, but I deleted it some commits ago.
18:49 chromatic ParrotInterpreter and ParrotThread look suspicious.
18:50 NotFound And partcl has some in non-singleton PMCs
18:50 dalek rakudo: d50cabc | masak++ | .gitignore:
18:50 dalek rakudo: [.gitignore] ignore some new generated files
18:50 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​50cabccae3c668f98fa92e187ccba5f0ec35b2b
19:00 gigabo left #parrot
19:03 dalek rakudo: 62f6082 | jnthn++ | src/classes/Object.pir:
19:03 dalek rakudo: Fix incorrect semantics in interaction of .? and multis.
19:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​2f6082e98b4c4e1a2d77eab1782152501fa0fe6
19:03 dalek rakudo: bae2cfa | jnthn++ | src/ (2 files):
19:03 dalek rakudo: A very first-cut implementation of lastcall, which just trims the candidate list such that nextsame and friends will find nothing else to defer to.
19:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/b​ae2cfa02728b37238b64bd964cdd216779281c5
19:05 Coke NotFound: removed.
19:05 szabgab joined #parrot
19:06 NotFound Coke: nice
19:09 Coke glad to help. now if you could just fix my enormous # of segfaults...
19:09 dalek partcl: r583 | coke++ | trunk/ (2 files):
19:09 dalek partcl: remove long-useless loadlibs
19:09 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=583
19:09 dalek partcl: r584 | coke++ | trunk/src/class/tclconst.pir:
19:09 dalek partcl: Remove stray debug method
19:09 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=584
19:09 dalek partcl: r585 | coke++ | trunk/tools/spec_info.pl:
19:09 dalek partcl: Allow git to work here as well as svn.
19:09 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=585
19:09 dalek partcl: r586 | coke++ | trunk/src/pmc/tcl (3 files):
19:09 dalek partcl: Remove now-unneeded if(pass) from PMCs.
19:09 dalek partcl: (Generated code is now smart enough to wrap this on our behalf)
19:09 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=586
19:09 dalek partcl: r587 | coke++ | trunk/runtime/builtin/foreach.pir:
19:09 dalek partcl: Add a hackity workaround.
19:09 dalek partcl: review: http://code.google.com/p/p​artcl/source/detail?r=587
19:10 Coke partcl segfaults is http://code.google.com/p/p​artcl/wiki/SpecTestStatus
19:11 NotFound Getting a copy of SpecTestStatus.wiki
19:11 NotFound Use of uninitialized value $repo in concatenation (.) or string at tools/tcl_test.pl line 56.
19:16 PacoLinux joined #parrot
19:26 Coke NotFound: you may be the first human to run that.
19:27 NotFound There is some prize?
19:27 Coke what does 'svn info .' say in your partcl checkout dir?
19:28 quek joined #parrot
19:28 NotFound URL: http://partcl.googlecode.com/svn/trunk
19:28 NotFound Repository Root: http://partcl.googlecode.com/svn
19:28 NotFound Repository UUID: 6cb8db7d-f34b-0410-8f57-4f83c6281724
19:28 NotFound Revision: 582
19:35 cotto chromatic, ping
19:37 rdice joined #parrot
19:37 joeri joined #parrot
19:37 gigabo joined #parrot
19:39 pyrimidine joined #parrot
19:39 Coke NotFound: looks like that should match the regexp on line 54, neh?
19:39 Coke what os?
19:41 chromatic pong
19:41 pyrimidine joined #parrot
19:41 cotto I'm trying to figure out the best strategy for dealing with PIR instructions that cause nested runl
19:41 cotto oops from a profiling point of view.
19:41 cotto The problem is that when an instruction causes a new runloop, all the instructions in that runloop are counted individually, but the original instruction gets all that timing info too.
19:41 cotto What do you think about having the post-processing script simply drop any instructions that cause n
19:41 cotto ested runloops?  It'd mean we'd lose a few (possibly important) instructions, but the timing information wouldn't be distorted.
19:43 dalek TT #919 created by jgabr++: On Mac OS X 10.5.8, Configure.pl : lazy symbol binding failed
19:43 chromatic Why does the original instruction get all that timing info?
19:43 chromatic Is that wallclock/system time or op count?
19:44 cotto system time
19:44 purl well, system time is 21 seconds off
19:44 cotto it all happens in DO_OP from the instruction's pov, since DO_OP creates the nested runloop
19:44 chromatic We could detect that though.
19:45 chromatic We know when we enter runops_profiling_core().
19:45 chromatic We know when we return from it.
19:47 cotto That wouldn't be completely accurate because it'd capture timing info for some extra code, but it'd be a decent approximation.
19:48 chromatic You can argue that it's more accurate in some ways.
19:48 theory joined #parrot
19:49 Coke no, system time is <reply>
19:49 purl okay, Coke.
19:51 cotto That'll certainly work.
19:55 NotFound Coke: Ubuntu i386
19:56 dalek TT #920 created by jgabr++: parrotbug should provide more guidance when it finishes
19:59 Coke NotFound: that should work, then.
19:59 Coke all it's doing is grabbing an svn export. not sure why that would fail on your platform.
20:09 Coke crap. well, thanks for trying to duplicate the segfaults, anyway. =-)
20:17 NotFound Coke: I'll take a look later.
20:33 Coke I appreciate it!
20:33 Coke NotFound++
20:33 Coke mikehh++
20:37 dalek TT #921 created by jgabr++: warnings in imcc/imcparser.c (statement with no effect)
20:38 chromatic I'm not sure we can fix TT #921, sadly.
20:40 Coke chromatic: we can by dropping IMCC and using pirc.
20:42 chromatic Well volunteered!
20:44 dalek parrot: r40548 | fperrad++ | trunk/compilers/tge/TGE.pir:
20:44 dalek parrot: [TGE] restore method find_key()
20:44 dalek parrot: which was removed from PGE in r40429.
20:44 dalek parrot: This method is useful for language which uses TGE.
20:45 dalek parrot: That's fix language Lua.
20:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40548/
20:48 quek left #parrot
20:49 Coke there's a rant about our deprecation policy for ya.
20:50 cotto Lua needs to get with the times.
20:50 cotto How hard can it be to reimplement a language that's already working?
20:51 dalek TT #851 reopened by coke++: [DEPRECATED]  .find_key method on PGE;Match objects
21:18 theory joined #parrot
21:22 Whiteknight joined #parrot
21:23 Tene Whiteknight: have you ever looked at memprof?
21:23 Tene http://www.rasterman.com/index.php?page=Memprof
21:23 Tene Whiteknight: Sorry I wasn't around last Sunday.  Life happened.
21:24 chromatic FOUL: Life doesn't happen to 23 year old techies.
21:25 Tene I can tell you what happened and provide video evidence for it if you'd like.
21:25 Whiteknight Tene: that's okay, I stayed plenty busy on Saturday
21:25 Tene I'm 23?  Lemme count.
21:25 Whiteknight Tene: actually I would like to see a video :)
21:25 Tene Ah, no, I'm 24 now.  I had a birthday.
21:26 chromatic Well then, nevermind!
21:28 Whiteknight Tene: when was your birthday?
21:28 Tene June 25.
21:36 cotto This is nice.  Now codingstds nits get fixed like as if by magic.
21:36 dalek parrot: r40549 | mikehh++ | trunk/compilers/tge/TGE.pir:
21:36 dalek parrot: fix codetest failure - pod syntax in compilers/tge/TGE.pir
21:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40549/
21:37 cotto Where "magic" is "mikehh".
21:40 darbelo joined #parrot
21:42 dalek rakudo: 4245753 | jnthn++ | src/ (3 files):
21:42 dalek rakudo: Make deferal failures into a kind of soft fail rather than an exception.
21:42 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​245753c82516b1a3f6c9f4f783392a8938914b4
21:42 dalek rakudo: a0082a8 | jnthn++ | t/spectest.data:
21:42 dalek rakudo: Add lastcall.t to spectest.data.
21:42 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​0082a8394342ca6f279f80b308ba85a2ce2444e
21:44 mikehh All tests PASS (pre/post-config, smolder, nqp_test, fulltest) at r40549 - Ubuntu 9.04 amd64 (gcc)
22:03 mikehh I started partcl make spectest and it seemed to hang and so I started rakudo tests and now am using about 100% of all 4 processors and partcl seems to started again
22:06 mikehh have started again
22:08 MoC joined #parrot
22:09 Coke mikehh: (hang) it could just be REALLY slow. see the last column for runtimes on feather.
22:09 Coke every time it starts a new .test file, it has to reparse all of library/tcltest/tcl*
22:12 kid51 joined #parrot
22:12 mikehh well rakudo tests PASSed and it back to 1 processor running at 100% for partcl- format tests now
22:13 mikehh rakudo (a0082a8) builds on parrot r40549 - make test/make spectest (up to 27994) PASS - Ubuntu 9.04 amd64
22:14 jonathan mikehh: At this rate, you won't need heating in winter...just to do enough spectest runs. ;-)
22:14 skv_ joined #parrot
22:14 mj41_ joined #parrot
22:15 allison joined #parrot
22:15 mikehh well temp is supposed to be 14C but it seems warmer in my room :-}
22:23 joeri left #parrot
22:28 pyrimidine left #parrot
22:30 TimToady joined #parrot
22:35 dalek rakudo: e5d0a37 | jnthn++ | src/parser/actions.pm:
22:35 dalek rakudo: Fix up code generation for hides trait_mod.
22:35 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​5d0a3793a8187090868f68f9c392cc985b4b393
22:35 dalek rakudo: 1a5c5f3 | jnthn++ | src/ (2 files):
22:35 dalek rakudo: Implement trait_mod:<hides> and a trait_mod:<is> for is hidden, plus add some attributes in ClassHOW that we'll use to track these.
22:35 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​a5c5f35a3963847784442a984439ac863655054
22:35 dalek rakudo: 637d803 | jnthn++ | src/ (2 files):
22:35 dalek rakudo: Make sure we also emit a call to trait_mod:<is>($class, :hidden), even though it also has some compiler side-effects (like suppressing *%_).
22:35 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​37d803b2bfbaf963ec2ba1515bf2c5bccc4e959
22:40 rg1 joined #parrot
22:45 mikehh well that was interesting - partcl make spectest took 46 min - not all tests reported (eg the first few)
22:46 mikehh I also built and tested rakudo while it was running
23:00 eternaleye Tene: Wow, your birthday is 2 days (plus a few years) before mine
23:00 Tene nice
23:00 dalek wmlscript: 053512d | fperrad++ | src/pmc/wmls (5 files):
23:00 dalek wmlscript: [cage] drop no longer useful "if (pass)" conditions from class_init functions
23:00 dalek wmlscript: See http://trac.parrot.org/parrot/ticket/918
23:00 dalek wmlscript: review: http://github.com/fperrad/wmlscript/commit​/053512d16afae9f117ef05e9b7d989245f44ee75
23:03 mikehh it seems for the partcl spectest the tests that did not report exited with - !! child died with signal 11, without coredump
23:03 cotto You'd think a bunch of computer geeks would be less surprised by the birthday paradox.
23:04 mikehh probability is one of the least intuitive areas
23:05 mikehh look what happens when you apply it to physics - as in quantum theory et al
23:06 cotto That doesn't seem likely.
23:06 cotto ;)
23:06 mikehh ;-}
23:29 Whiteknight joined #parrot
23:40 eternaleye cotto: It's more surprising when you consider that there were >4 people who had birthdays this week - strictly speaking, that's a rather higher number of incidences that the norm
23:41 eternaleye (considering the number of people in the channel)
23:41 cotto I'll leave that as an exercise to the reader.
23:49 dalek parrot: r40550 | cotto++ | branches/pluggable_runcore (2 files):
23:49 dalek parrot: [profiling] first attempt to deal gracefully with nested runloops
23:49 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40550/
23:50 kthakore joined #parrot
23:54 kid51 Uh oh!  Smolder is down again !
23:54 nperez joined #parrot

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

Parrot | source cross referenced