Post closed tickets in your report. | Note: This channel is for our weekly status meetings (Tuesdays at 19:30 UTC); you probably want #parrot instead. | irclog: irclog.perlgeek.de/
Set by moderator on 6 September 2011.
01:06 whiteknight joined 05:13 contingencyplan joined 08:06 wagle joined 10:28 Coke joined 12:33 bluescreen joined 13:10 Coke joined 13:50 contingencyplan joined 13:57 bluescreen joined 15:00 Coke joined 16:29 eternaleye joined 16:35 bluescreen joined 16:44 whiteknight joined
whiteknight WHAT I DID: 16:52
* Lots of talk about new policy stuff.
* Played with parrot-libgit2 a little bit, fixed up some old Rosella-based test stuff
* Big refactor for Rosella Harness library. Am working to improve test coverage for it all now. Internals are much prettier now and more extendable
* Adding several new features to Rosella Query and FileSystem. Need to blog about them and document
* Added try/catch/throw to Jaesop, with tests
* Looking at Parrot GC stuff. Sent an email to parrot-dev about it. Want to rip out some of the run-time core selection stuff and optimize some hotpaths.
WHAT I WILL DO:
* Wrap up changes to Rosella Harness
* Start a branch for exploratory GC optimizations
* Talk about merging whiteknight/kill_threads, then kill all thread-related tickets.
* More looking at 6model. Probably want to talk to jnthn at some point
* Knock off a few more to-do items in Jaesop, get stage 0 "complete"
EOR
17:00 HSlayer joined
Coke What I DID: started hacking on a smolder replacement. Collaborators (or even just users) welcome. EOR. 17:27
17:42 dukeleto joined 18:28 NotFound joined
NotFound What I did: 18:39
-parrot
* Minor improvements and cage cleaning in frontend/parrot2
* Fixed Timer repeating usage
* Cage cleaning and fixes in scheduler
* Updated winxed snapshot to get latest features and fixes
-winxed
* Fixed HLL issues in class operator
* Don't automatically tailcall by default insside a try block
* allowtaicall modifier in try, from whiteknight++ and plobsing++ comments
* builtin __ASSERT__
* -X command line option fron whiteknight++ Issue 11
What I will do:
No plan
EOR
19:09 Coke joined
cotto_work #ps in 8 19:22
19:25 tcurtis joined
tcurtis DONE: nothing, just relaxed. WILL DO: Start working on LALRskate again. EOR 19:25
cotto_work *did: 19:29
- chat with whiteknight about the Next Big Thing
-- probably C/6model import into Parrot
-- also, GC macroization
-- it's still changing in nqp, so there's that
- merged dafrito++'s pull request to update Pod::Simple
- overally, it's been relatively quiet
- misc subprof hacking
*todo
- subprof cleanup, grokking, possibly stealing the good parts
- finish profiling docs
*eor
Util # Nothing done, and no plan, due to $WORK. 7-day ticket report:
3 closed: fixed
2 closed: invalid
2 new
.end
cotto_work hello 19:30
Util Hello 19:31
bubaflub hello 19:32
tcurtis Hi. 19:34
cotto_work How's this week been? I like the cleanup work that's been happening.
dukeleto ~~ 19:35
NotFound Hola
dukeleto we have a few very basic tests for hll interop in t/pir/interop.t now 19:36
cotto_work I saw. dukeleto++
not enough though. ;)
dukeleto cotto_work: should that test become part of "make test" ?
cotto_work: currently, it is only part of "make interop_tests"
cotto_work dukeleto: I'm disappointed that it's not. Why shouldn't it?
dukeleto cotto_work: no good reason. Mostly, i just forgot to add it 19:37
cotto_work wfm. Add that sucker.
dukeleto cotto_work: i imagine those tests will, for example: attempt to load Lua and Rakudo pbc. If they exists, attempt to marshall data between Lua + Rakudo, etc...
19:38 benabik joined
dukeleto cotto_work: in theory, we could test interop between winxed and nqp without needing anything outside the parrot repo 19:38
cotto_work dukeleto: I'm a bit apprehensive about having tests like that. I want interop to be well-tested, but having tests run/pass conditionally on whether hlls are installed strieks me as suboptimal' 19:39
(referring to Rakudo and Lua)
dukeleto cotto_work: no, they would be skipped if the necessary HLL's are not present
cotto_work right, so they'd usually get ignoref
*ignored
dukeleto cotto_work: yes. What is a better solution? Where should HLL interop tests live? Their own special repo? 19:40
benabik If they're part of make test, they will generally get run before any HLLs have been installed/updated.
cotto_work Hmmm. allhlltest pretty much guarantees that HLLs will be available. That might be a starting point.
dukeleto benabik: well, winxed and nqp will be present, at least
whiteknight Jaesop isn't at a point where it can really be tested 19:41
it was waiting on some HLL fixes that were just added to Winxed 19:42
cotto_work Its day will come. We'll be ready.
whiteknight that day might have to wait for 6model. There is no reasonable way to do HLL mapping with javascript objects until we have much more subclassable metaobjects 19:43
cotto_work btw, have you been looking at jnthn++'s 6model changes in nqp over the past ~week?
whiteknight I skimmed a few. Need to look harder 19:44
cotto_work Is there any kind of well-defined process for bringing in changes or do we just as jnthn to sit still? ;)
whiteknight I'm sort of in a holding pattern right now
dukeleto i am sure our current way to do hll interop will change, but having some kind of basic tests for it seems useful. For instance, it uses P6metaclass right now, which will most probably change. 19:45
whiteknight I'll be getting back to that this week. I really need to coordinate with jnthn about how we want to work this
cotto_work btw, anyone even remotely interested in 6model is strongly encouraged to study the code. We need as high a bus number on that as we can get.
benabik I've looked it over, but wouldn't say I'm expert in it.
cotto_work same here 19:46
dukeleto cotto_work: i was reading all the commits for a few months, but haven't looked at it recently
cotto_work: is the plan to pull in 6model wholesale, with possibly a few modifications, or rewrite it? 19:47
that is not clear to me
NotFound I'm ready to add features to winxed compreg'ed object, as long as they are going to be useful. If nqp or rakudo are going to expect some metaclass on it, will be a lose of time.
whiteknight my current plan is to pull it wholesale
Coke q1q
whiteknight that is, copy what is in NQP right now into Parrot, so NQP feels as little change initially as possible
Once it's in, we start integrating. 19:48
cotto_work I'd prefer a rewrite so that we're not trying to adapt code for a situation that it's not designed for.
whiteknight whether that's the best plan or not is open to debate. I'm open to comments
cotto_work I'll need to spend some quality time with the code anyway.
whiteknight we can do it either way. A rewrite will be more intensive of developer time
cotto_work (if it turns out that the code won't require much adaptation, I'm happy with that approach too) 19:49
benabik The 6model core seems fairly lightweight and clear. We probably won't want to import NQP's HOWs, but it seems to me that the basic STABLE, attributes, API, etc. shouldn't need much change.
(We'd replace NQPClassHOW with a PMCHOW or something, I'd imagine.) 19:50
cotto_work Does anyone have other thoughts before we move to coke's question? 19:51
benabik has a class starting in 10, so will probably go AFK in a moment.
cotto_work Coke: go ahead. 19:52
until Coke gets back, does anyone else have a question? 19:56
I guess that's it for questions. 20:00
We had a few goals set from last week. 20:01
GOAL 1: work with Rakudo on a new support policy
GOAL 2: filter through feedback on support policy, keep track of the important bits
GOAL 3: document allhlltest (git_workflow.pod)
GOAL 4: new SSL cert
GOAL 5: figure out a rough roadmap of what we want to change (whiteknight)
whiteknight roger
cotto_work The first goal depends on Rakudo being ready. I'll talk to them as ask them when they think that'll happen. 20:02
The second goal is similar. I don't think we have anything more organized than the parrot-dev thread. 20:03
Util FYI, #phasers is moving to @1730UTC, skipping tommorow. 20:04
Wednesdays, I should have said
cotto_work I'll add allhlltest to git_workflow now. 20:05
anyone know about the ssl cert? 20:06
whiteknight I sent an email to OSU, they said we needed some kind of key. I pinged Jerry about it
cotto_work alright 20:07
I know there's a gist of possible goals for a post-deprecation Parrot floating around somewhere. any objections if I create a "misc" repository under the parrot org for that kind of thing? 20:11
dukeleto cotto_work: more repos are always nice, but perhaps a better name than "misc" ? 20:12
cotto_work let the bikeshedding begin! 20:13
benabik Blue!
dukeleto Clear with polka dots.
cotto_work: i guess "misc" is fine :)
cotto_work I'm open to a better name. 20:14
benabik gh-pages? :-D 20:15
dukeleto cotto_work: what are you planning in putting in this repo? 20:16
cotto_work: mostly textual documents for planning?
cotto_work dukeleto: that and anything else that should be generally useful, shared and doesn't really fit elsewhere
shared rw, that is 20:17
dukeleto cotto_work: "misc" is fine for now. If we think of a better name, we can rename it. 20:18
cotto_work dukeleto: this is true 20:19
20:22 soh_cah_toa joined
cotto_work created 20:23
whiteknight: can you stick your roadmap gist in there? 20:24
whiteknight yeah, I will tonight
cotto_work thanks
any goals for the week?
soh_cah_toa submit smolder reports to catch test failures for upcoming release 20:25
also, get whiteknight/kill_threads merged into master
whiteknight if there are no objections, I think we want to do that
cotto_work +1 20:26
dukeleto +1 to kill threads
whiteknight I think we're starting to get enough interest built up behind a replacement 20:27
Util +1 20:28
whiteknight has to sign out now.
cotto_work anything else? 20:31
or should we call it a wrap
benabik +1
cotto_work It sounds like we're done. Go forth and learn 6model! 20:32
and make sure not to destabilize the release 20:33
20:37 benabik left
Coke returns, far too late for his question. 20:37
cotto_work ooc, what was it? 20:38
Coke -> #parrot 20:40
20:45 soh_cah_toa left 21:39 NotFound left 23:01 bubaflub joined 23:52 whiteknight joined