Perl 6 - the future is here, just unevenly distributed

IRC log for #parrotsketch, 2010-05-18

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

All times shown according to UTC.

Time Nick Message
00:48 particle joined #parrotsketch
01:37 bakkdoor joined #parrotsketch
02:04 bluescreen joined #parrotsketch
06:39 eternaleye joined #parrotsketch
06:49 eternaleye joined #parrotsketch
09:18 bakkdoor joined #parrotsketch
11:55 bluescreen joined #parrotsketch
12:24 bluescreen joined #parrotsketch
12:37 Whiteknight joined #parrotsketch
14:03 eternaleye joined #parrotsketch
15:41 bluescreen joined #parrotsketch
15:48 Coke Rakudo - - Working on some sample code for masak.
15:48 Coke Parrot - - fix editor/skeleton.pir - fix crow
15:48 Coke (er, paste failure.)
15:48 Coke Rakudo - - Working on some sample code for masak.
15:49 Coke Parrot - - fix editor/skeleton.pir - fix crow - add some experimental/deprecated tickets to track status.
15:49 Coke Partcl -
15:49 Coke - Trying to get it passing make test again -
15:49 Coke - Find a parrot segfault. (fixed by notfound++ & plobsing++)
15:49 Coke - However, currently failing all tests. Could use more eyes.
15:49 Coke WillDo: - Begin preparing for 2.6 stable release. - rip out recursive make for docs/ - more dependency fixups (and autogeneration)
15:49 Coke .EOHORRIBLYFORMATTEDREPORT>
15:55 bluescreen joined #parrotsketch
15:55 Whiteknight joined #parrotsketch
15:55 spinclad joined #parrotsketch
15:55 pmichaud joined #parrotsketch
15:55 Util joined #parrotsketch
15:55 Tene joined #parrotsketch
15:55 cotto joined #parrotsketch
15:55 wagle joined #parrotsketch
16:10 bluescreen joined #parrotsketch
16:10 Whiteknight joined #parrotsketch
16:10 spinclad joined #parrotsketch
16:10 pmichaud joined #parrotsketch
16:10 Util joined #parrotsketch
16:10 Tene joined #parrotsketch
16:10 cotto joined #parrotsketch
16:10 wagle joined #parrotsketch
17:28 mikehh joined #parrotsketch
17:56 mikehh What I did since my last report:
17:56 mikehh * building and testing parrot on amd64/i386, with gcc/g++
17:56 mikehh * various fixes
17:56 mikehh * branch testing and some fixes
17:56 mikehh * fixed perlcritic in ops_pct branch to ignore files with shebang line of #! nqp
17:56 mikehh What I intend to do in the next week:
17:56 mikehh * testing and fixing
17:56 mikehh * documentation
17:56 mikehh .eor
18:04 NotFound joined #parrotsketch
18:06 NotFound What I did:
18:06 NotFound -parrot
18:06 NotFound * Added find_codepoint opcode following pmichaud++ suggestion and
18:06 NotFound discussion on IRC.
18:06 NotFound * Added experimental function Parrot_str_unescape_string, intended
18:06 NotFound to sanitize semantic of string literals in pir.
18:06 NotFound * Shortened generated code for the Object PMC.
18:06 NotFound * Fixed and updated pbc_checker.
18:06 NotFound * Added unroll opcode to handle the inner runloop problem in exception
18:06 NotFound handling.
18:06 NotFound * Miscellaneous fixes and cleanups.
18:06 NotFound -winxed
18:06 NotFound * Always escape non-ascii chars in generated PIR string literals.
18:06 NotFound * Use the unroll opcode.
18:06 NotFound * Use pir .const directive for const string.
18:06 NotFound What I will do:
18:06 NotFound No plan
18:07 NotFound EOR
18:15 cotto_work joined #parrotsketch
18:43 cotto_work #did:
18:43 cotto_work - opsc cleanup work
18:43 cotto_work - made ops2c.nqp use Getopt::Obj for arg processing
18:43 cotto_work - switched ops_pct to use ops2c nqp fakecutable to build ops by default
18:43 cotto_work - moved ops.h code out of ops2c.nqp into opsc
18:43 cotto_work - nuked ops2c perl code in ops_pct branch
18:43 cotto_work - updated opsc tests
18:43 cotto_work - found a bug introduced by NotFound++, who quickly fixed it and added a test
18:44 cotto_work - figured out opsrenumber
18:44 cotto_work - grammar
18:44 cotto_work - fixed most or all misuses of "it's" in svn
18:44 cotto_work - you're welcome
18:44 cotto_work - gsoc
18:44 cotto_work - met with Khairul on Monday, answered some technical questions (extra runcore data and dynpmc building)
18:44 cotto_work - he'll be working in a svn branch as soon as he gets his commit bit
18:44 cotto_work #will do:
18:44 cotto_work - move opsrenumber functionality into opsc
18:44 cotto_work - answer any questions khairul has, do code review
18:44 cotto_work #eor
18:44 cotto_work q1q (and I bet you can guess what it is)
18:45 tcurtis joined #parrotsketch
18:56 bluescreen joined #parrotsketch
18:58 darbelo joined #parrotsketch
19:03 chromatic joined #parrotsketch
19:04 khairul joined #parrotsketch
19:14 Whiteknight * Did:
19:14 Whiteknight - lots of testing
19:15 Whiteknight - cut the 2.4.0 "Sulfur Crest" release
19:15 Whiteknight *EOR
19:25 Chandon joined #parrotsketch
19:36 tcurtis What I've Done:
19:36 tcurtis * Wrote a blog post about the API of PAST::Pattern and related things.
19:36 tcurtis * Started work on an implementation of Go for Parrot(named Fly, since that's how Parrots go) then stopped upon realizing that all the PCT docs are for PGE and old NQP.
19:36 tcurtis * Asked for a low-level PMC role for things that support invoke VTABLE and got "invokable" thanks to bacek++
19:36 tcurtis * Finished high school.
19:36 tcurtis What I will do:
19:36 tcurtis * Officially graduate high school
19:36 tcurtis * Start coding on PAST::Walker.
19:36 tcurtis * Hopefully get a commit bit
19:36 tcurtis EOR
19:50 bubaflub joined #parrotsketch
20:00 tcurtis UNEOR the url of mentioned blog post is http://www.parrot.org/content/past-pattern-matching EORAGAIN
20:00 chromatic I ... didn't get much done and I likely won't until the weekend at the earliest.
20:02 darbelo DONE
20:02 darbelo -    Build cleanups in ops_pct branch.
20:02 darbelo -    Started implementing ucs-4 expect commits by tomorrow.
20:02 darbelo -    Finishing up this week's blog post right now.
20:02 darbelo TODO
20:02 darbelo -    Finish ucs-4. Turned out ICU isn't as helpful there as I hoped.
20:02 darbelo -    Finish the blog post. Half an hour tops.
20:02 darbelo -    Create an svn branch and start cracking on the grapheme table implementation.
20:02 darbelo EOR.
20:09 Util # Done:
20:09 Util * Created branch pbc_to_exe_direct_output to reduce memory usage for Rakudo build
20:09 Util = The memory issue was already alleviated by strings improvements in trunk, but this removes the hot-spot for the future.
20:09 Util * Recovered my work on TT#1302 (PIR todo() is frequently misused).
20:09 Util = Commiting tomorrow.
20:09 Util # Plan to do:
20:09 Util * Brain-dump pbc_to_exe knowledge into a Wiki page.
20:10 Util # Blockers:
20:10 Util * $WORK
20:10 Util .end
20:22 allison joined #parrotsketch
20:26 allison Between the Ubuntu Developer Summit and preparing for my first exam tomorrow morning, I had no time to work on Parrot this week.
20:26 allison EOR
20:30 cotto_work good afternoon
20:30 smash joined #parrotsketch
20:31 mikehh hi all
20:31 chromatic Hello, everyone.
20:31 Util Hello
20:31 darbelo Hola.
20:31 smash hello everyone
20:31 allison hiya
20:31 NotFound Hola.
20:31 chromatic Let's review last week.
20:31 chromatic 2.4.0 is out.  How'd that go?
20:32 moritz joined #parrotsketch
20:32 Coke whiteknight isn't here.
20:32 Coke he didn't complain much. more announcements to go out later. had to back out some last minute commits that came in too close to the release.
20:32 darbelo We had a slight hitch with the trunk freeze, IIUC.
20:33 chromatic Any thoughts on preventative measures?
20:33 cotto_work talk to fperrad
20:34 Coke a git workflow might help avoid it.
20:34 darbelo We might want to use svn ACL features in the future, as a stop-gap before git time is here.
20:34 Coke keep an eye on commit log emails if you're the RM is a good temporary workaround.
20:34 darbelo That too.
20:35 chromatic Other thoughts?
20:35 cotto_work Wouldn't non-technical measures be a better first resort?
20:35 moritz I don't think it happens often enough to require thorough investigation yet
20:35 darbelo cotto_work: Probably, but I'm no good with those.
20:36 chromatic Let's start with non-technical stuff.
20:37 chromatic Other review of last week.  How'd we do on tests and documentation?
20:38 chromatic That doesn't sound promising.
20:39 mikehh some tests were added and I did some testing
20:39 NotFound I added some tests after breaking things because of not having such tests X-)
20:39 Util I am working on wiki braindumps of pbc_to_exe, and of the build process in overview, but have nothing posted yet.
20:40 chromatic I know Util's looking at a review of some TODO tests.
20:40 chromatic Are there other places where we have potential breakage due to the lack of tests?
20:41 Util Mostly those are just TODO changing to SKIP, or to conditioned TODOs; it will not improved them beyond that.
20:41 darbelo At least one of the codetests was taught stay out of ext/
20:42 darbelo That's an improvement.
20:42 mikehh there are some TODO's that pass on i386 but not on amd64 (Ubuntu that is)
20:42 chromatic We should make that a design priority, to identify undertested systems when we're working on them.
20:42 chromatic That goes double for systems that are currently difficult to test.
20:43 Util re: potential breakage - should we be running the test suite on installed_parrot?
20:43 chromatic +1
20:44 allison yes
20:44 mikehh AFAIK there are problems there
20:44 allison yes, it doesn't work at the moment
20:45 allison but it *should* work
20:45 allison useful for testing packages after installing them too
20:45 chromatic Any volunteers to look into it?
20:45 NotFound That TODO is flawed, it uses a file build in a particular architecture file for an arch dependant test.
20:46 mikehh I'll look at it - how far I'll get is another story
20:46 chromatic Knowing the problems is itself good progress.
20:46 chromatic Anything else to review from last week?
20:47 NotFound We need to build some pbc with the current parrot for that kind of tests, instead of using native_pbc
20:47 mikehh we got a lot of stuff related to codestring moved to trunk
20:47 mikehh string_builder
20:47 Coke that's all in StringBuilder.
20:47 Coke right.
20:47 NotFound I added the unroll opcode, review and opinions needed.
20:48 NotFound TT #1635
20:48 chromatic allison and I talked about syntax improvements to PIR to do something similar.
20:50 NotFound Probably a helper syntax will need the opcode anyway.
20:50 chromatic I'd like to experiment with that, especially with NQP.
20:50 chromatic I'm curious to see how it affects Rakudo.
20:51 NotFound I'm using it in winxed right now, but don't make specific testing yet.
20:52 chromatic Let's move on to plans for this week.
20:52 chromatic What should we focus on?
20:52 mikehh ops_pct merge?
20:53 cotto_work should be possible
20:53 * bacek is there... Almost.
20:53 cotto_work just needs opsrenumber
20:54 mikehh how almost is that
20:54 cotto_work well, that's more iwbn than a need
20:55 chromatic Sounds like that's well under way.  Any other priorities?  Rakudo needs?
20:56 darbelo jnthn reported failures to the list. Any news on that?
20:56 chromatic Looked related to ICU versions to me.
20:57 darbelo Anything our config step could do about it?
20:57 chromatic If it's related to ICU 3 versus ICU 4, config could refuse to use ICU older than 4.x
20:58 chromatic I don't know that we've verified if it's that, though.
20:58 cotto_work quick unrelated q, if I may
20:58 Coke jnthn couldn't duplicate it on his laptop but only his desktop, so it's something versiony.
20:58 darbelo Also, we should probably revisit if we want to keep ICU optional or make it mandatory.
20:59 cotto_work I'd like to propose that my gsoc student khairul (trac username "khairul") be given a commit bit with me as his mentor.
20:59 darbelo Not right now, but maybe before our next supported release.
20:59 chromatic When does coding start?
20:59 darbelo The 24th
21:00 allison darbelo: last pass through we decided to say that ICU is required for Rakudo, but optional for Parrot
21:00 allison (to avoid unnecessarily heavy installs of Parrot in resource constrained systems, like, say, mobile phones)
21:00 cotto_work afk 20m
21:01 darbelo allison: Then rakudo needs to enforce it. I've built rakudo on ICU-less parrots and only noticed after tests failed for me.
21:01 Coke darbelo: that's up to rakudo.
21:01 darbelo I'll look into it later and see if I can send them a patch for it.
21:02 allison darbelo: should be easy enough for rakudo to check that in their config
21:02 Coke good luck. pretty sure jnthn isn't using icu. =-)
21:03 Coke If folks who opened deprecation tickets can make sure they have a summary that says "This is what the new thing looks like", it would be helpful for those trying to close out these old tickets.
21:04 chromatic Any other thoughts on ICU and versions and potential failures, or do we need more data?
21:04 tcurtis moritz is collecting version number data on that, I think.
21:05 chromatic Alright.
21:05 chromatic Let's move on to GSoC.  Do students need commit bits?
21:07 Coke I see no problem with giving them bits for the project duration assuming mentors are about.
21:07 Coke (and have already done that for one I think.)
21:07 mikehh if they are used responsibly
21:08 chromatic Standard lecture: work in your branch, don't break trunk, be responsible.
21:08 allison also fine with it, especially if they're primarily working on branches
21:08 allison (and usual CLA rules)
21:08 chromatic Students in the channel: please contact your mentors.  Mentors, please shepherd your students.
21:09 Whiteknight joined #parrotsketch
21:11 chromatic Shall we move on to roadmap review?
21:12 bacek +1
21:12 chromatic allison?
21:13 allison no work on GC this week
21:13 allison that's pretty much it
21:13 chromatic I'm blocked on time... and waiting to hear about bacek's fixed-number-of-gcables bitmask marking approach.
21:15 bacek chromatic, it's still "sweep" phase. May be little bit faster due reading only bitmask instead of full headers.
21:15 chromatic Can you write up a description of the algorithm and data structures on the wiki?
21:16 bacek chromatic, yes, later today.
21:16 chromatic I hope to be able to start implementing something after this weekend.
21:16 chromatic Any other questions?
21:17 chromatic Let's wrap it up then.
21:17 bacek Can we deprecate gc_mark_pobj_alive?
21:17 chromatic In favor of what?
21:17 bacek In favour of PObj* gc_walk_pobj(PObj**)
21:18 chromatic For copying/compacting?
21:18 bacek Yes
21:18 chromatic I'm not certain that's the right approach.  It seems early to do that.
21:19 bacek Basically, there is only one place where we have information about PMC memory layout - VTABLE_mark
21:20 bacek And I do want to extend it to provide more information for GC
21:21 bacek (Let's discuss it on #parrot)
21:22 chromatic Okay, thanks everyone.
21:23 tcurtis I have a question.
21:23 chromatic Go ahead.
21:23 tcurtis Is there any up-to-date documentation for PCT?
21:24 tcurtis If not, can someone who is knowledegable about it work on some?
21:24 Whiteknight may need to write it
21:24 chromatic pmichaud keeps threatening to work on it, but it's definitely not there.
21:25 bacek tcurtis, perldoc compilers/pct/src/PAST/Nodes.pir
21:25 Coke tcurtis: they told me the same thing when I asked. there is no good tutorial documentation, methinks.
21:27 tcurtis bacek: I was insufficiently specific. I was inquiring about overall documentation from a compiler-writer's perspective, especially concerning grammars.
21:29 Tene tcurtis: the documentation right now is: the PCT source, and the source of other compilers that use PCT.
21:29 bacek tcurtis, I don't think that we have (good) docs for nqp-rx. You can read squaak tutorial for slightly outdated usage of PGE/PCT.
21:30 * Util has had to clone-and-hack existing NQP, for lack of good NQP docs.
21:31 Coke bacek: that's most unhelpful. =-)
21:32 bacek Coke, at least it covers PCT part :)
21:33 Tene I'd love to work on PCT docs.  It's been on my tasklist for a long time.
21:33 Coke tene++
21:51 chromatic MEETING OVER
21:52 NotFound left #parrotsketch
21:52 Util so noted :)
22:13 cotto_work left #parrotsketch
22:21 darbelo left #parrotsketch
22:33 Chandon left #parrotsketch
23:43 eternaleye joined #parrotsketch

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