"Tuesday at 20:30 UTC"
Set by moderator on 20 April 2010.
04:31 tcurtis joined 06:14 japhb joined 12:08 bluescreen joined 14:33 plobsing joined
plobsing might not make #ps, nopasted report at "Tuesday at 20:30 UTC" 14:34
07:23 -!- plobsing [~pitr@199.247.190.36] has joined #parrotsketch
07:23 -!- Topic for #parrotsketch: "Tuesday at 20:30 UTC"
07:23 -!- Topic set by Coke [~coke@cpe-72-228-52-192.nycap.res.rr.com] [Tue Apr 20 11:58:37 2010]
07:23 [Users #parrotsketch]
07:23 [ ascent ] [ dukeleto ] [ japhb ] [ pmichaud] [ TimToady] 14:35
07:23 [ bluescreen] [ eternaleye] [ particle] [ sorear ] [ Util ]
07:23 [ Coke ] [ ilbot2 ] [ PerlJam ] [ spinclad] [ wagle ]
07:23 [ cotto ] [ integral ] [ plobsing] [ tewk ]
07:23 -!- Irssi: #parrotsketch: Total of 19 nicks [0 ops, 0 halfops, 0 voices, 19 normal]
07:23 -!- Channel #parrotsketch created Sun Sep 18 22:54:01 2005
07:23 -!- Irssi: Join to #parrotsketch was synced in 1 secs
oops.
trying again.
might not make #ps, nopasted report at nopaste.snit.ch/20528
particle better!
Coke Here is plobsing's inlined report: 14:57
What I Did:
* fixed performance of string reverse lookup in IMCC (with help from bacek++)
* closed tickets (1209, 1244, 1073, 1030, 1152, 1201, 1035, 1570, 162)
What I Plan (not in any particular order, not all to be done immediately):
* IMCC
* use PMCs internally
* restructure to not use interp as a god object
* syntax for constants (would fix some instances of broken PASM)
* more dead code elimination
* more ticket/bug fixes
* NCI 14:58
* struct PMCs
* move libjit_framebuilder to github
Questions:
* tickets requiring advice: (1323, 1610, 132)
END_REPORT
(again, that was plobsing's report, not mine.)
15:36 cotto_work joined 16:28 bakkdoor joined 16:54 darbelo joined 17:03 PacoLinux joined 17:04 PacoLinux left
cotto_work #did: 17:39
* helped bacek++ make progress on ops_pct
- fixed dynop support, Parrot's internal dynops currently pass all tests
- opsc (nqp ops compiler) still needs to be installed and integrated with existing build tools
- opsc appears to cause no new test failures when used to build Rakudo's dynops
- making it work without an installed opsc is a bit hacky though
#will do:
* make opsc installable, improve documentation, get ops_pct ready to merge (before 2.4 is possible, before 2.5 is certain)
#closed TTs:
#eor
17:42 Tene joined 18:47 mikehh joined 19:10 plobsing_work joined 19:11 smash joined
Coke Did: 19:17
* Got (old) partcl to compile against parrot-trunk. Segfaults immediately.
* poked at branches/codestring - trunk got faster faster than we did, so
unless we get some better benchmarks soon, we probably need to consider
it a failed experiment.
* added some text to trac.parrot.org/languages so folks trying to checkout
the source can find it.
* fixed more make dependencies.
EOR
mikehh What I did since my last report: 19:22
* building and testing parrot on amd64/i386, with gcc/g++
* codetest and ASSERT_ARGS fixes
* branch testing and some fixes
* ops_pct branch passes all tests except perlcritic fails on 9 nqp files
What I intend to do in the next week:
* testing and fixing
* look at ways to incorpotate nqp-rx files into codetest properly,
* particularly geting perlcritic to ignore them, but other tests as well
* documentation
.eor
19:34 bubaflub joined
darbelo DONE 19:35
- Finished up with the various forms submissions for Google.
- Removed #if 0 code from various places.
- Added some code to prevent the "*_provisional" Config keys from ending up in config_lib.pir
- Did some experimenting for future string cleanups.
- Read and re-read PDD28.
- Blogged some more about NFG.
TODO
- Keep up the blogging.
- Clean up strings/io a bit more.
- Maybe start with the NFG coding.
END.
20:00 chromatic joined 20:01 NotFound joined
NotFound 2010-may-11 20:05
What I did:
-parrot
* Update and fix example pbc_checker.
* Improve memory usage in packfile annotations pack/unpack.
* Improve PackfileAnnotations PMC.
* Sanitize Parrot_str_length, Parrot_str_byte_length and its
corresponfing opcodes.
* Added experimental function Parrot_str_unescape_string.
* Minor fixes and cleanups.
-winxed
* Escape non-ascii chars in generated PIR strings in stage 1
What I will do:
No plan
EOR
Tene What I did: 20:07
* Compiled and installed Parrot and Rakudo on my new laptop, made vague allusions to possibly working again someday
KTHXBAI
20:08 tcurtis joined 20:10 khairul joined
Util # Done: 20:20
* Investigated pbc_to_exe speed issues on 64-bit Linux
# Plan to do:
* Answer outstanding issue on TT#1302 (PIR todo() is frequently misused).
= ++bubaflub for code effort, ++kid51 for recovering my dropped ball.
* Refactor pbc_to_exe for direct-to-filehandle emit of c-encoded PBC
* *or* add a Linux-only codepath to pbc_to_exe for direct linking of PBC
= (just like Win32 .res files; `ld -r -b binary -o foo.o foo.pbc`)
= this is a GCC-only solution; will not work with Darwin's linker.
* Brain-dump pbc_to_exe knowledge into a Wiki page.
# Blockers:
* $WORK
.end
20:21 allison joined
tcurtis What I did: 20:22
* Worked on the design of my GSoc project some. 20:23
* Posted a blog post about PAST::Walker at parrot.org/content/past-optimization
* Submitted my CLA(actually did this the week before this last one, but I missed last #ps).
What I will do:
* Work on the design of PAST::Pattern or whatever I'll call it.
* Post a blog post about PAST::Pattern
* Finish high-school so that I'll have more time to work on Parrot.
* Possibly start coding.
End of report.
chromatic I... am blocking on work. 20:24
allison What I did:
- Took a bit of a tangent this week and wrote a PEG parser. www.bitbucket.org/allison/spix. An experiment partly inspired by working on GC, reading up on PEG parsers, and a stray comment from chromatic about the number of objects created by PGE/NQP-rx. It's like a simplified PGE, enough for parsing computer languages, but not the full power of regular expressions.
- At the Ubuntu Developer Summit this week.
EOR
chromatic Hello, everyone. 20:32
cotto_work hi
Util Hello
mikehh hello
allison hi
NotFound Hola
japhb o/ 20:33
chromatic Let's review last week's milestones. 20:34
Deprecations?
Branch merges? 20:36
Coke not as many this week. 20:37
branches/codestring wasn't merged. needs more benchmarking, but it looks like trunk leapfrogged us.
cotto_work ops_pct will be ready for merging soon 20:38
Coke I think cotto & bacek are doing a final push on the ops_pct rewrite.
chromatic What happens after ops_pct?
mikehh codestring and ops_pct mostly test ok except for perlcritic in ops_pct
allison do we have some branches that are no longer active?
chromatic Off the top of my head, the tt389 and avl_cache 20:39
allison some of them are 7-11 months since the last change
plobsing_work libjit_framebuilder2 is inactive. I intent to mine the useful stuff out of it and then delete it relatively soon. 20:41
Tene Didn't we almost get tt389 done?
allison Tene: it wasn't the right fix 20:42
Tene Ah.
allison the right fix went into trunk already 20:43
but, we kept the branch for a bit, just in case something useful could be pulled from it
chromatic There's a lot of cleanup we need in that code.
We have a 2.4 release in a week. What should we work on in this week? 20:44
cotto_work sounds like a good weekly task
allison clean up the namespace PMC?
maybe the week after the release? 20:45
how about ticket sprint or documentation?
chromatic I'm all for a ticket sprint.
allison sounds good 20:47
chromatic Roadmap review?
allison How do we split GC into smaller chunks?
chromatic I don't think we do. 20:48
darbelo Smaller code chunks?
allison smaller tasks
chromatic I don't know a way to break up the separate marking system for sweep-free such that you can replace it piecemeal. 20:49
allison Or, how do we make it an interesting thing to work on.
the dogpile approach has worked well in the past
darbelo Get someone qualified to do the big changes and the farm out the smaller fixed to the community? 20:50
allison basically, how do we get people to work on it
chromatic The dogpile last week was speeding up Rakudo and dropping its memory usage after a couple of branch merges.
allison that was good
Util Document/braindump everything about GC, current design and ultimate thoughts. Some tasks on research, benchmarking, and instrumenting will probably come clear during the writing.
allison we just added the 3-month roadmap type of task in the last development summit 20:51
so, we're still working out the practical details of how it integrates with our daily development
Util: that's a good starting point 20:52
I've been thinking about doing a parrot.org blog post
chromatic I can put some tasks on the sweep-free page.
allison there are also tasks on the GC page 20:53
Is the best way to get people working on them to just dive in myself?
chromatic I think the GC scares almost everyone.
allison yeah, I think you're right 20:54
okay, that's my usual job of clearing away nastyness to make the way for others
I can handle that
end of roadmap review
chromatic If you and bacek can fix up the documentation on how to make a new GC core, that's a good start.
allison that's a nice concrete starting point, I'll work on that and a GC blog post this week 20:55
chromatic Other questions, concerns, or discussion items? 20:57
allison git migration volunteers
chromatic Timeline? 20:58
allison I figure find the volunteers first and let them discuss the timeline.
chromatic cotto's already looking at Git/Trac. 20:59
allison there are some intertwined factors
chromatic dukeleto has volunteered to do training.
darbelo After Rakudo Star, maybe after GSoC, form minimal disruption.
allison well, I didn't want to volunteer them, but yes, that's who I suspected
chromatic After GSoC is the earliest I like too.
allison and, that gives us time to plan our dev practices on the new system 21:00
do we have an updated timeline for rakudo star? 21:01
chromatic My guess is by YAPC::NA.
allison okay, so before GSoC is over anyway
do we have a volunteer to write the revised development practices? 21:03
chromatic What are you thinking needs revision? 21:04
allison well, the currently give the best practices for svn 21:05
committer_guide.pod, branching_guide.pod, metacommitter_guide.pod 21:06
chromatic Those'll probably need more discussion. 21:07
allison release_manager_guide.pod
right, what I'm looking for is someone willing to drive the discussion
raise the questions, resolve them in a new form of the document
Coke dukeleto and I can do that. 21:09
allison awesome, thanks!
Coke (he said, volunteering him.) 21:10
allison do we have a wiki page yet for the git migration?
chromatic Looks like a no. 21:12
allison creating one now and adding those roles 21:14
end of q 21:15
chromatic Anything else to discuss today?
mikehh I was looking at native_pbc last week - those tests have been skipped for at least 15 months (TT #357) - anything being done there?
also we have signals currently disabled, anything there? 21:16
chromatic Not to my knowledge.
mikehh should we be doing something? 21:17
chromatic I think signals depend somewhat on concurrency. 21:18
Portable bytecode is a ways off.
allison which is to say, yes, but not immediately 21:19
chromatic I expect post-Lorito bytecode to get a lot more portable. 21:21
allison we'll at least have fewer ops to port 21:23
Tene I'm also available for handling git training/questions/headaches at any time. 21:29
allison thanks
chromatic Anything else? 21:31
mikehh I was also going to bring up shebang line consistancy, but I will leave that until I investigate further
21:33 bacek joined 21:41 Whiteknight joined
mikehh hey bacek, Whiteknight, you have been volunteered for ... 21:41
Whiteknight ...? 21:42
Whiteknight starts backlogging
cotto_work The meeting seems to be over. 21:54
Coke we kind of petered out there. See everyone next week. =-) 21:56
allison trac.parrot.org/parrot/wiki/GitMigration 21:59
22:01 darbelo left 22:44 NotFound left

Powered by Cro and the Raku® Programming Language.

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

Thank you!