Vision for 2.0: Production Users | Priority for 1.6: Merge Branches | Priority for this week: Fix partcl segfaults & PANICs, increase test coverage on Namespace and Array PMCs, prune a struct | trac.parrot.org/parrot/wiki/Propos...chProtocol | Note: This channel is only for our Tuesday status meetings; you probably want #parrot instead.
Set by moderator on 6 September 2009.
05:02 ilbot2 joined
moderator Vision for 2.0: Production Users | Priority for 1.6: Merge Branches | Priority for this week: Fix partcl segfaults & PANICs, increase test coverage on Namespace and Array PMCs, prune a struct | trac.parrot.org/parrot/wiki/Propos...chProtocol | Note: This channel is only for our Tuesday status meetings; you probably want #parrot instead.
08:14 allison joined
allison - Reviewed objects PDD, and created a short list of remaining tasks. This closes the roadmap milestone for pdd15. 08:14
- Reviewed multiple inputs on the JIT and put together a plan of attack.
- Talked with Whiteknight and chromatic about exception handlers and inferior runloops.
- Unblocked infinite looping test failures on the pcc branch, started on named parameter failures. Plan to make a fresh branch for that work from trunk once the old JIT is disabled in trunk (may wait for the old code removal too).
- Preparing for talks at the JVM Languages Summit, and Bay Area Python user group.
- Unfortunately I'll miss #parrotsketch this week, will be in the air over Canada.
09:43 mikehh joined 11:42 mikehh_ joined 11:45 mikehh joined 13:57 Whiteknight joined
particle this week: released parrot 1.6.0. thanks to all who helped this go so smoothly. 13:58
next week: release rakudo #21.
no blockers atm.
Whiteknight What I did last week: 14:47
* More research on JIT, reading LLVM Documentation, etc
* Tested building Parrot with Clang. Doesn't currently build with GMP, need to track down the reason for that
* Looked into the Inferior Runloops problem, talked about solutions with NotFound++, Allison++, chromatic++
* Lots of testing for the release, but no real fixing/hacking
* Wrote up a Y-Combinator implementation in PIR, very useful for testing/learning about the newclosure opcode (which is not well documented)
* Busy with $real_life mostly 14:48
What I plan this week:
* Kill JIT. Kill it dead.
* Kill PIC. Kill it with fire.
What I am blocking on:
* Family issues in $real_life
cotto # What I did: 15:30
* got a nice optimization to the profiling runcore
- more than 4x faster for oofib.pir, improvement will be proportional to the number of context switches
- chromatic++ for giving me the functions to work with
# What I hope to do and how many tuits I expect to have:
* get started on testing the profiling runcore
* apply an optimization to pprof2cg suggested by bacek++
* start resurrecting ops_pct for the coming jitpocalypse
* employment has somehow lessened my tuit supply
# What could block my progress:
* rl
eor
mikehh What I did: 16:29
* building and full testing parrot on amd64 and i386
* fixing codetest/manifest_tests failures
* testing language builds mainly rakudo and partcl
What I want to do:
* continue testing
* add some more tests
.eor
16:45 darbelo joined
darbelo # What I did: 16:45
* Got a commit bit. Broke the build.
* Removed some broken code in pmc_freeze.
* Did a minor cleanup there, too.
* Removed some more strstart uses. 16:46
#What I want to do:
* Make pmc_freeze use the string API.
* Remove strstart for good, eventually.
eor
16:50 cotto_work joined 17:06 smash joined 17:26 cotto_work joined 17:27 pmichaud joined
pmichaud What I did: 17:34
* Had lots of non-Parrot/Rakudo/Perl6 tasks this week
* Rakudo now passing 15,500 tests, an increase of 3,131 tests since
the 2009-08 release
* Re-calculated size of spectest suite (using more accurate algorithms
from moritz++) for 2009-08 release to present
* Enabled HLL mapping for Float->Num in Rakudo.
** This fixes a number of tests, and only shows a slight speed penalty
in spectests on my platform.
** Other platforms may be seeing speed issues from this change, reports welcome.
What I'm doing this week:
* More of what I had planned for this last week
* Writing up more documentation for PCT, NQP, Rakudo Star planning
* Adding contextual variables to NQP
* Working on many other PGE/NQP updates
* Designing protoregexes for PGE
* Have $otherjob tasks this week so won't get well into it until Saturday
What I'm blocking on: 17:35
* Insufficient usable hours per solar day
* $otherjob
EOR
17:36 Util joined 17:53 NotFound joined
NotFound What I did: 18:05
* Helped and being helped for diagnosing bugs on irc.
* Fixing, fixing, fixing...
* Applied patch from einstein++ that fixes a nasty GC bug
* Verified that after an NCI fix MySql.pir works again
* Updated xlibtest.p6 and mysqltest.p6 to current rakudo syntax
* Fixed MySql pir for usage from rakudo, unfortunately not in time for 1.6.0
* Applied some other patches
What I will do:
* No fixed plan ATM. Maybe improve MySql.pir
EOR
18:06 davidfetter joined 18:13 mikehh_ joined 18:19 mikehh joined
Util # Done: 18:24
* Upgraded to Mac OS 10.5, enabling GCC4, GitX, Valgrind, and fixing my long-broken ICU.
* Verified as now working: Packfile PMCs can roundtrip a basic hello.pbc (TT#504).
# Plan for next week:
* Make SVN properties do-the-right-thing for `git-svn` users (TT#994).
# Blockers:
* None; more tuits available this week.
.end
Whiteknight Hello 18:31
darbelo Hola
18:32 chromatic joined, flh joined
mikehh hello 18:32
Util Hello
particle io
chromatic Hello. 18:33
NotFound hola
Whiteknight allison isn't here today
particle silently sings "oh, canada"
chromatic Let's review last week's tasks. 18:34
How did we do fixing segfaults, panics, and odd bugs?
Whiteknight I think we nailed it
NotFound Some were fixed, some workarounded
Whiteknight and by "we" I mean not me
darbelo Coke looked pleased on his mail.
Whiteknight ..and I've never seen Coke look "pleased" 18:35
chromatic How confident are people?
Whiteknight very
mikehh looks good
NotFound I thinke now we are better than in 1.5.0 and before
chromatic We still have to design and implement our way out of exception handlers and inferior runloops, but the bugs I saw NotFound et al patch were nasty, subtle ones. 18:36
cotto_work hi
chromatic How did PMC testing improvements go?
NotFound Maybe we must add some functions that doesn't require casting to PObj, to better detect such errors 18:37
Whiteknight I didn't do anything for testing this week
chromatic Where's duk3leto, he should know. 18:38
NotFound I think FPA improved again, but doesn't verified it.
chromatic I'm sure we didn't get any struct pruning done. I spent most of my time diagnosing and fixing weird bugs. 18:39
Whiteknight ...and there were lots of weird bugs!
particle and that was more important.
chromatic Yes.
mikehh the coverage on the tests we targetted improved
pmichaud agreed... this week saw lots of good work
particle thanks to everyone for the good work this month, the release was extremely easy, smooth, and boring.
chromatic particle, would you do the honors of migrating 1.6 roadmap items? 18:40
particle my only frustration was how long it took to do the press.
NotFound Forgot to mention in report: added an attribute on Continuation to help detect reuse errors
particle chromatic: i'm not sure what belongs where. were any completed, or do they all move forward?
chromatic I closed the one I know we completed (review Objects PDD).
pmichaud mine were not completed
chromatic I was going to ask. 18:41
Util I have info to report on the Packfile PMC milestone
My only remaining issue with Packfile PMCs is that you cannot use them to build a .pbc from scratch. (which is needed to start making actual use of them in refactoring existing Parrot tools)
In TT#504 (comment 17), bacek says that this is not due to any shortcoming of Packfile PMC; supporting code is missing from elsewhere in Parrot.
I say, we call it "landed", and fork off a non-milestone ticket for the "can't build from scratch" issue.
chromatic +1 18:42
Whiteknight +1
chromatic Let's talk about 1.7 milestones. 18:43
vtable swap -- I care about this one a lot less now with the Key refactor. I think we can deprioritize it.
particle util, care to close out tt#504 then?
Util particle: I will.
chromatic duk3leto, are you here? If so, how does the debugger documentation look? 18:44
cotto_work q1q 18:45
chromatic The other priority is a bytecode testing framework. Anyone interested in that?
Okay, let's leave that one as is. 18:46
What should we work on this week, besides removing the JIT? 18:47
mikehh I'll have a look at it - but don't guarentee anything
darbelo Increase test coverage for something else? 18:48
chromatic How is the coverage for Array now?
Whiteknight I had a question about Array: what is it's purpose? 18:49
mikehh much better - on the fixed stuff anyway
Whiteknight is it substantially different then RPA?
chromatic It's an optimization, but not a big one. 18:50
duk3leto 'ello. sorry I'm late.
chromatic: The debugger docs still need improvement.
chromatic Okay, it's a milestone for 1.7 duk3leto. 18:51
Whiteknight I would suggest that if RPA and Array are the same that we only need one or the other
at least, the same in terms of functionality
pmichaud they aren't the same 18:52
Whiteknight okay, then that answers my question
pmichaud at least, there's an inheritance involved
Array -> FPA -> RPA
cotto_work pmichaud, nothing inherits from Array.
particle FPA inherits?
Array is on its own 18:53
Whiteknight pmichaud: so is there some kind of capability that RPA has that Array doesnt?
pmichaud oops
NotFound pmclass PackedArray extends Array {
particle iirc it stores native int/float/pmc, not boxed
pmichaud you're right, I got my inheritance mixed up (or it changed)
NotFound That's in docs/vtables.pod
particle it never changed
Array and PackedArray are experimental as far as i'm concerned, they're not widely used and not well maintained 18:54
NotFound Oh, is just an example
chromatic Last week's suggesting for testing was Exception and MultiSub.
Whiteknight I like those 18:55
duk3leto chromatic: i think we should stay with those for this week
cotto_work PackedArray is currently fictional or historical
pmichaud I think I'm fine with deprecating Array
Whiteknight we're going to want lots of tests for Exceptions if we're going to be refactoring that system
NotFound Is just an example, we can change it to MyOwnArray to avoid confusions 18:56
chromatic Other priorities for this week?
mikehh we need to merge any branches in the next couple of weeks 18:57
Whiteknight I don't think there are many of those this month
particle JIT
NotFound If we are going to change the way we handle it, write more tests in the current way is a loss of time
chromatic Only the PCC branch sounds outstanding.
mikehh leave the last two weeks to stabalize
duk3leto this is a good time to do deprecations. It gives rakudo lots of time to deal with them. 18:58
particle ripping out jit should happen this week, i say
it unblocks future development
duk3leto particle: if not today
darbelo Does anyone know the status of jrtayloriv's gc-refactor branch?
duk3leto who wants the pleasure to rip out JIT? is there a branch?
particle there is a branch 18:59
pmichaud kill_jit
particle parrot: r41277 | darbelo++ | branches/kill_jit:
Whiteknight gc-refactor is good and probably ready to merge (thanks for the reminder)
although jrtayloriv is without net access for a while
chromatic Okay. Question time. 19:00
cotto?
darbelo q1q
cotto_work Is it our long-term plan to require that any custom aggregate PMC have its own dedicated iterator and iterator key PMCs?
pmichaud (if anyone is looking for a ticket to work on, TT #389 is a big issue for Rakudo)
cotto_work i.e. Hash, HashIterator, HashIterator, HashIteratorKey 19:01
particle doesn't see any answers. ask on the mailing list. 19:03
cotto_work will do
chromatic darbelo?
cotto_work eoq
cotto_work looks forward to another warnocking
darbelo Is there any policy on strct member removal when they affect languages? 19:04
NotFound I think the policy must be: don't use struct members 19:05
chromatic The deprecation policy excludes them from backwards compatibility support. 19:06
darbelo ok. 19:07
eoq
chromatic Other questions? 19:08
pmichaud a reminder that trac.parrot.org/parrot/report/16 has a list of current HLL issues that need parrot help 19:09
chromatic Thanks everyone; good work for 1.6. 19:10
19:12 pmichaud left 19:13 Util left 19:16 darbelo left 19:17 NotFound left 19:35 mikehh_ joined 20:08 mikehh joined 20:42 cotto_work joined 21:05 Whiteknight joined 23:05 PacoLinux left 23:47 duk3leto_ joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!