Camelia, the Perl 6 bug

IRC log for #parrotsketch, 2010-06-08

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

All times shown according to UTC.

Time Nick Message
00:26 tcurtis joined #parrotsketch
04:35 integral joined #parrotsketch
04:51 tcurtis_ joined #parrotsketch
05:13 eternaleye joined #parrotsketch
11:00 whiteknight joined #parrotsketch
15:45 contingencyplan joined #parrotsketch
16:42 tcurtis joined #parrotsketch
17:00 NotFound joined #parrotsketch
17:09 cotto_work joined #parrotsketch
17:13 NotFound What I did:
17:13 NotFound -parrot
17:13 NotFound * Added a bunch of tests, near 100% coverage of ...Array PMCs.
17:13 NotFound * Fixed some tests that were failing because of dynops, not of the feature
17:13 NotFound being tested.
17:13 NotFound * Modify some PMC vtable functions to reduce complexity, simplifying its
17:13 NotFound coverage.
17:13 NotFound * Deleted do-nothing custom mark in StringBuilder.
17:13 NotFound -winxed
17:13 NotFound * Minor refactors.
17:14 NotFound * Predef constants true and false.
17:14 NotFound * Improved example ajax.
17:14 NotFound What I will do:
17:14 NotFound No plan
17:14 NotFound q2q
17:14 NotFound EOR
17:23 Coke Done:
17:23 Coke - resurrected partcl
17:23 Coke - Sneakiest breakage was result of TT #389 fix - you can no longer
17:23 Coke put subs or their children into a namespace unless you declare them
17:23 Coke in PIR and add the :nsentry flag. This means you cannot DYNAMICALLY
17:23 Coke without the PIR compreg. Aooga.
17:23 Coke Will Do:
17:23 Coke - Now, partcl-nqp is broken. something is eating a CONTROL_RETURN exception
17:23 Coke when it didn't used to.
17:23 Coke EOR
17:23 Coke *DYNAMICALLY /add/ them.
17:29 bubaflub joined #parrotsketch
17:46 whiteknight joined #parrotsketch
17:50 tcurtis_ joined #parrotsketch
18:27 darbelo joined #parrotsketch
18:37 mikehh joined #parrotsketch
19:04 chromatic joined #parrotsketch
19:05 chromatic I read some commits and replied to some messages and that's really all.
19:05 PacoLinux joined #parrotsketch
19:52 khairul joined #parrotsketch
19:53 allison joined #parrotsketch
20:00 allison My time this week was still absorbed by exams. Finished the last one yesterday, so will have time again starting this week.
20:00 allison EOR
20:06 pmichaud + Fixed up 'exit' opcode, added CONTROL_EXIT exception type.
20:06 pmichaud + Modified PAST so that it can generate symbolic pasm constants in PIR output.
20:06 pmichaud + Applied bacek++ patches, nqp-rx now supports multisubs and multimethods
20:06 pmichaud + Fixed sigspace handling on nqp-rx ** quantifier in regexes
20:06 pmichaud + Added \e to nqp strings
20:06 pmichaud + Worked out use of inversion lists for charclass lists in regexes
20:06 pmichaud + Fixed some substr-out-of-range errors in Regex.match
20:06 pmichaud EOR
20:13 cotto_work khairul, ping
20:13 mikehh What I did since my last report:
20:13 mikehh * building and testing parrot on amd64/i386, with gcc/g++
20:13 mikehh * various fixes
20:13 cotto_work ww
20:13 mikehh * branch testing and some fixes
20:13 mikehh What I intend to do in the next week:
20:13 mikehh * testing and fixing
20:13 mikehh .eor
20:18 cotto_work #did:
20:18 cotto_work - helped khairul with some questions
20:18 cotto_work - moved bacek's linked list code in gc_massacre into a place that's more amenable to reuse
20:18 cotto_work - pinged OSU about getting a trac/git test site; no reply
20:18 cotto_work - worked on LoritoRoadmap
20:18 cotto_work #will do:
20:18 cotto_work - get trac/git set up
20:19 cotto_work #blockers:
20:19 cotto_work - OSUOSL responsiveness
20:19 smash joined #parrotsketch
20:19 cotto_work #eor
20:19 cotto_work q3q
20:19 Util # Done:
20:19 Util * Edited Perl6book.
20:19 Util # Plan to do:
20:19 Util * Write Perl6book section on currying.
20:19 Util * TT#1302 (PIR todo() is frequently misused).
20:19 Util # Blockers:
20:19 Util * $WORK
20:19 Util .end
20:22 darbelo DONE
20:22 darbelo -    Various cleanups and some added functionality to NFG. Various parts of the STRING API are now more graceful in their handling of grapheme tables. concat still isn't.
20:22 darbelo -    Got myself a windows dev environment for my birthday. In retrospect, I should have gotten some teeth pulled...
20:22 darbelo -    Some windows testing.
20:22 darbelo TODO
20:22 darbelo -    Stay on schedule.
20:22 darbelo -    Tackle concatenation and COW issues.
20:22 darbelo -    Blog post for this week is coming later today.
20:22 darbelo EOR.
20:26 khairul Done:
20:26 khairul - Had some fun exploring dynop loading process (http://parrot.mangkok.com/?p=94)
20:26 khairul - Added some tests for the Instrument dynpmc.
20:26 khairul - Made codetest slightly happier.
20:26 khairul Will Do:
20:26 khairul - Finish up reworking runtime library.
20:26 khairul - Write synopsis for week 2.
20:26 khairul - Explore making loading dynops a stop the world event like GC.
20:26 khairul EOR
20:27 tcurtis What I did:
20:27 tcurtis * Mostly implemented PAST::Pattern.
20:27 tcurtis - Can match on types, children, and attributes,
20:27 tcurtis - Can match based on iseq to a constant,
20:27 tcurtis - Can match based on true result from a closure, and
20:27 tcurtis - Can match based on anything with an ACCEPTS method.
20:27 tcurtis - Can't match globally yet.
20:27 tcurtis - Can't transform a PAST where it matches yet.
20:27 tcurtis * Forgot to post a blog post last week.
20:27 tcurtis - Posted one yesterday.
20:28 tcurtis What I will do:
20:28 tcurtis * Finish implementing PAST::Pattern.
20:28 tcurtis * Start writing optimizations using PAST::Pattern.
20:28 tcurtis * Not forgot my blog post this week.
20:28 tcurtis * Write some documentation.
20:28 tcurtis * Start working on adding optimizations to PCT compilers.
20:28 tcurtis * Also start working on some example optimizations.
20:32 bubaflub Done:
20:32 bubaflub * Almost finished Parrot configure on RTEMS
20:32 bubaflub Will Do:
20:32 bubaflub * Switch to use perl or autoconf for the last step
20:32 bubaflub Blockers:
20:34 bubaflub repaste:Blockers:
20:34 bubaflub * Work, Makefile limitaitons
20:34 bubaflub EOR
20:34 chromatic Anyone else?
20:35 chromatic Let's begin then.
20:35 chromatic How did we do on last week's goals?
20:35 dukeleto Did:
20:35 dukeleto * Worked on PL/Parrot and PL/Perl6
20:35 dukeleto * GSoC stuff
20:35 dukeleto Will do:
20:35 dukeleto * GSoC stuff
20:35 dukeleto Blocking on:
20:35 dukeleto * Unable to load perl6.pbc from PIR: http://trac.parrot.org/parrot/ticket/1675
20:36 dukeleto .EOR
20:36 * dukeleto says hello
20:36 dukeleto what were the goals?
20:37 NotFound Hola
20:37 darbelo Hi.
20:37 chromatic Dynops breakage, for one.
20:38 chromatic The gc_massacre branch for two.
20:38 chromatic What's the status of dynops and trunk and clean test runs?
20:40 mikehh there is only 1/4 failing test in make corevm/make coretest now - make corevm/make coretest - t/pmc/io.t - Failed tests:  25, 27, 29, 31
20:41 mikehh it's related to loading io/ops
20:41 chromatic Who's working on them?
20:42 cotto_work plobsing last I checked
20:43 chromatic Any objections to keeping that as a priority for this week?
20:43 chromatic (as we have a release next week)
20:44 NotFound chromatic: Fixing the tests or fixing the problem?
20:44 chromatic Ideally the latter.
20:44 NotFound +100
20:44 cotto_work I don't know if it's solvable by then
20:45 particle i suggest updating tutorials/docs as a priority this week, we're preparing for the R* release
20:45 particle squaak is broken
20:46 mikehh the error in all the subtests  - error:imcc:loadlib directive could not find library `io_ops'
20:46 chromatic Do we have a volunteer to coordinate fixing the IO problem/tests?
20:46 mikehh which I think is in dynopslib which has not been built yet in make corevm
20:48 mikehh I'll be working on it - I think it might involve moving the subtests out of t/pmc into dynopslib or something
20:48 chromatic Okay.
20:48 chromatic Anyone else for reviewing and updating documentation?
20:49 Util +1
20:49 NotFound mikehh: I agree, there is no login in testing some dynops in t/pmc
20:49 NotFound s/login/logic
20:50 mikehh if I have time I'll work on documentation as well
20:50 chromatic Okay, any other suggestions for the week?
20:51 Util Does Rakudo have any special needs from Parrot, in prep for R* ?
20:51 NotFound There was some talk about the need for a ByteBuffer or something like that.
20:53 cotto_work They're working on their own dynpmc.  It may later be replaced by a yet-unwritten core PMC.
20:53 pmichaud I don't believe we have any specialized needs for R*.
20:53 pmichaud We may be adding a few past-related items in the next 3-4 days, but nothing that should affect core PMCs or functionality
20:54 NotFound I like the idea of that PMC, I can write it this week.
20:54 chromatic Last chance to get bug requests in....
20:54 pmichaud at the moment I don't know of any bugs that are blockers for us
20:54 pmichaud the one I found this past week (exit opcode) was easily fixed, and if we had to we could've worked around it rather than fixing it
20:54 NotFound Someone has a better name than ByteBuffer?
20:54 pmichaud most of the bugs seem to be in that category these days
20:55 mikehh the last time I tested rakudo against trunk - it passed all tests - just some TODO's passing (earlier today)
20:55 bacek joined #parrotsketch
20:55 * bacek yawns
20:56 chromatic Alright, it sounds like we're on track then.
20:56 chromatic Let's go to questions.
20:56 mikehh bacek: too early for you :-}
20:56 chromatic NotFound?
20:56 pmichaud I'll run a final test on saturday-ish and raise alarm bells if I think of anything.
20:56 NotFound chromatic: one has been answered by allison on #parrot, the other is the ByteBuffer
20:57 chromatic Do you know enough to make a test implementation?
20:57 NotFound I think so.
20:58 chromatic Okay, cotto_work has three questions.
20:58 cotto_work What's the eta on the gc_massacre merge?  I'd like to get the linked list code into trunk for khairul soonish.
20:58 bacek cotto_work, not before after 2.5...
20:59 mikehh got to pass all tests first - it hangs on me
20:59 bacek But we can cherry-pick list code into trunk.
20:59 cotto_work ok.  In that case I may end up trying to get the list code into trunk without a full merge.
20:59 chromatic Does khairul need it before next Wednesday?
21:00 cotto_work nafaik
21:00 khairul no its no hurry.
21:00 cotto_work I'll wait for it to be merged then.
21:00 cotto_work eoq1
21:00 cotto_work to people interested in Lorito: is there anything obviously missing from http://trac.parrot.org/parrot/w​iki/LoritoRoadmap#LoritoDesign , especially from the primary issues?
21:02 spinclad (send answers back to #parrot?)
21:02 cotto_work wfm
21:03 chromatic Last question?
21:03 cotto_work next q:
21:03 cotto_work I want to get Lorito's design process started in earnest.  The issues that need to be resolved are listed in approximate order of depencence on LoritoRoadmap.
21:03 cotto_work I'd like to get volunteers to start wiki pages similar to the PDDs.  Once these are somewhate complete, we can use them as a starting point for further discussion about how Lorito will work.
21:03 cotto_work any volunteers?
21:04 chromatic I can take on a couple.
21:04 cotto_work ok.  which ones?
21:05 chromatic purpose and balance
21:05 whiteknight joined #parrotsketch
21:05 cotto_work anyone else?
21:06 allison I'll help
21:06 allison I haven't looked at details yet, so won't put my name on a specific task yet
21:08 cotto_work Thanks.  More volunteers are welcome.
21:08 cotto_work eoq
21:08 mikehh cotto_work: It is definately something I am interested in
21:09 chromatic Any other questions?
21:10 chromatic Anything else?  Roadmap review?
21:10 mikehh Just want to check on the policy regarding what is in corevm regarding dynops/dynoplibs
21:10 cotto_work That'd be good to clarify.
21:11 allison I'm out of touch on the GC work
21:11 allison so, the gc_massacre update is the roadmap review
21:12 chromatic Needs polish.
21:12 chromatic Perhaps a tiny bit faster than trunk.
21:12 NotFound Forgot one question.
21:12 bacek A lot of polish.
21:14 mikehh string bits need implementation
21:14 NotFound The printerr problem. Now we don't have any mean to write to stderr without loading dynops other that the stdhandle method, wich is experimental.
21:15 Tene I'd rather bring in getstderr than printerr.
21:15 allison agreed
21:15 dukeleto q1q
21:16 bacek I did bring getstd* opcodes back already
21:17 NotFound bacek: yeah, but I ask what is the desirable long term solution.
21:17 Tene Then can't you call the 'print' method on the result of getstderr?
21:17 allison NotFound: $P0 = getstderr; print $P0, "foo"
21:18 NotFound So the get/set std are going to be in core permanently?
21:19 allison yes
21:19 NotFound eoq then
21:19 chromatic dukeleto
21:20 dukeleto chromatic: yes! my question:
21:21 dukeleto Does anybody know what is going on with my error on loading perl6.pbc ?
21:22 dukeleto Also, is anyone against getting rid of the open/close opcodes after 2.6 is released? Can I add them to DEPRECATED?
21:22 Tene shouldn't you be using load_lang for that, or was that abandoned with the other hll api changes?
21:22 NotFound dukeleto: +1
21:23 dukeleto http://trac.parrot.org/parrot/ticket/1675 is the error I get when loading perl6.pbc from PIR and from C
21:23 chromatic Sounds like some difference between :onload and :init somewhere in perl6.pbc.
21:24 dukeleto chromatic: do you have any idea about how "parrot perl6.pbc" is different? Because that actually works. It may have something to do with packfiles
21:26 chromatic I can never remember which of :onload or :init happens when or under which circumstances, but I bet pmichaud could give a guess about what perl6.pbc does.
21:27 chromatic This makes me want to remove both, enforce that packfiles have some sort of "When I get compiled!" method, and pass in some sort of "Here's the context of how you were compiled!" object for them to do what they want.
21:27 tcurtis :load is "Run this subroutine when loaded by the load_bytecode op", :init is "Run the subroutine when the program is run directly" according ot PDD19.
21:28 chromatic If my suspicion is correct, you should be able to produce the same error with parrot some_file.pir which loads perl6.pbc inside.
21:30 chromatic Anything else?
21:32 chromatic Let's call it a week then.  Fix bugs, everyone.
21:32 bubaflub left #parrotsketch
21:35 darbelo left #parrotsketch
21:36 NotFound left #parrotsketch
21:41 PacoLinux left #parrotsketch
22:20 eternaleye joined #parrotsketch
22:40 cotto_work left #parrotsketch
23:31 eternaleye joined #parrotsketch

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