Vision for 2.0: Production Users | Don't forget to post closed tickets in your report. | Note: This channel is only for our Tuesday status meetings; you probably want #parrot instead. | irclog: irc.pugscode.org/
Set by moderator on 2 February 2010.
00:31 wagle joined, pmichaud joined, cotto_work joined, Tene joined, tewk joined 06:29 cottoo joined 08:25 eternaleye joined 12:56 bluescreen joined 14:36 particle joined 15:09 PacoLinux joined 15:34 particle joined 16:00 particle joined 16:34 whiteknight joined 16:37 particle joined 18:28 mikehh joined
whiteknight WHAT I DID: 18:56
* Wrote some blog posts about the state of the exceptions system, after hearing some complaints and reading the docs.
* Fixed Parrot_range_rand to be inclusive of the upper-bound after the issue was identified by Austin
* Fixed some build warnings for Coke, including one which required input from chromatic
* Got commit bit to Kakapo, read over a lot of the code and added a few test cases for super()
* Made a proposal about a new VTABLE that would allow delegation of method lookup to a metaobject, in addition to find_method (which delegates lookup to the invocant). Wrote to the list about it, largely with no direct responses.
* Looked into some issues involving Objects, subclassing, MRO, and vtable overrides, starting to see some common threads in multiple tickets involving these topics.
* Created the new fix_icc_failures branch to fix some of the nagging arithmetic test failures when Parrot builds with ICC. dukeleto and a few other intrepid hackers have downloaded the compiler and may be helping with the work.
* Updated some tickets and answered questions on IRC.
WHAT I WILL DO:
* Would like to put together an experimental branch with the proposed new find_class_method VTABLE to show off it's merit
* Planning for a pure-PIR PIR/PASM compiler library. Highly inspired by NotFound's recent work on winxed. A basic PASM compiler should be relatively simple to do, if performance is not an obstacle.
* More digging into the object metamodel, have some things I want to test out and maybe get fixed.
WHAT I AM BLOCKING ON:
* Would like more feedback on my VTABLE_find_class_method idea.
19:22 cotto_work joined 19:39 cotto_work joined 19:41 NotFound joined 19:55 dukeleto joined 19:56 cotto_w0rk joined 20:00 chromatic joined
chromatic Fixed a couple of bugs; limited time last week. That will likely continue this week. 20:00
20:01 Coke joined
mikehh What I did since my last report: 20:04
* building and testing parrot on amd64/i386, with gcc/g++
* branch testing
* got t/codingstd/perlcritic.t to PASS (with help from kid51++)
* codetest now PASSes again
What I intend to do in the next week:
* testing and fixing
* looking at cleaning up some tests
* documentation
.eor
20:05 cotto_work joined
NotFound What I did: 20:05
- parrot
* Tested a imcc patch that avoids duplicating STRING constants
with the same value/charset/encoding.
* New PMC flag manual_attrs
- parrot and Winxed:
* More work towards HLL written PIR compilers, adding a few methods
in the Opcode and OpLib PMC and using in Winxed example 'pirado'.
What I will do:
* No plan
EOR
Coke branches/rm_cflags/ 20:07
- now warnings-clean on GCC.
- still needs optimize fix (some compilers cannot optimize some files some times). Figure about one more week before this gets done and merged back.
trunk/
- move SDL::* into SDL;* and other leftovers in runtime/parrot/library
- fix TT#1402 - trying to find 'buc' in 'bubuc' would fail, as one of the 20:08
iterators was never reset. mls++ for some speed updates on top of that.
- fix check for PIR and Perl files when searching t/
.
whoops: also: changed trac.parrot.org/parrot/report/4 to show all assigned tickets, not just ones in the state of "assigned". (some folks own new tickets) 20:11
Tene * Worked on exceptions refactor. Making good progress there (for me) 20:18
* Ran into confusing issues with subclasses and attributes. Implemented a workaround for now, but still not sure... Emailed the list, but no responses.
* Probably going to continue working on exceptions refactor this week.
* Considering working on rakudo or other HLLs, but I seem to be very busy lately.
KTHXBAI
20:18 bacek joined 20:19 kurahaupo joined
kurahaupo dukeleto: I was expecting #parrotsketch @ 18h30Z; has it changed? 20:20
20:21 cotto_w0rk joined
Tene kurahaupo: people pre-post for several reasons. 20:21
then we don't have to go through coordinating and waiting for the reports at the start of the meeting.
cotto_w0rk It's a very nice change from how we used to run the meeting. 20:22
bacek Done: 20:23
* More work on boehm_gc_2 branch. Stuck for now due some shenanigans in freeze/thaw.
* Resurrected ops_pct branch.
Todo:
* Finish end-to-end skeleton of ops_pct.
EOR.
Util # Done:
* Worked more with Amazon EC2, in preparation for a talk at local LUG.
= EC2 looks like a excellent tool for testing CPAN on Perl5-on-Parrot, when the time comes.
* No Parrot time; no Parrot progress.
# Plan to do:
* Submit talks to YAPC::NA. 20:24
# Blockers:
* $WORK
.end
cotto_work # did:
* removed some unused and barely-used ops2c code
* got back into opsc hacking now that bacek++ has it working again
# will do:
* opsc stuff
#blockers
* self
#eor
* the end
dukeleto What I did: * Got accepted to give a talk about PL/Parrot with David Fetter at PGCon 2010 (20-21 May 2010 at University of Ottawa) * Fiddled with PL/Parrot * Fiddling with compiling Parrot with clang and icc 20:26
What I will do: * Prepare for GSoC stuff. If you are interested in being a mentor for Parrot, please ping me. * Get rid of some warnings on Parrot with clang + icc
Blocking on: * 42
EOF
lets try that again 20:27
What I did: * Got accepted to give a talk about PL/Parrot with David Fetter at PGCon 2010 (20-21 May 2010 at University of Ottawa) * Fiddled with PL/Parrot * Fiddling with compiling Parrot with clang and icc
What I will do: * Prepare for GSoC stuff. If you are interested in being a mentor for Parrot, please ping me. * Get rid of some warnings on Parrot with clang + icc
Blocking on: * 42
EOF
dukeleto fails
20:27 bubaflub joined 20:32 cotto_mibbit joined
bacek o hai 20:32
NotFound Hola
whiteknight hello
Util Hello
mikehh hello 20:33
cotto_mibbit hi
bubaflub hello
dukeleto 'ello
chromatic Hello.
Let's review last week's tasks.
How goes the PCC refactor?
whiteknight nil
20:34 cotto_w0rk joined
bacek looks like no one touched it 20:34
chromatic Any theories on what's blocking it?
whiteknight ENOALLISON 20:35
chromatic Is there a volunteer to talk to her and figure out how to coordinate work on it?
whiteknight I'd be happy to, if it can be done wthout her
chromatic I think it can, but I don't think anyone knows quite what to do. 20:36
I've read the wiki page and I didn't know where to start.
bacek It can be done. But my scars from previous PCC refactoring still fresh.
whiteknight I have a vague idea, but more input would be nice
chromatic Okay, Whiteknight gets deputized to break things into tasks and dole them out.
Any other volunteers to help with that branch? 20:37
whiteknight done.
bacek whiteknight, I can give you some input for PCC refactor.
mikehh testing and that type of stuff
whiteknight bacek: awesome. in #parrot
chromatic HLL subclassing and MMD... I fixed one bug and could use help with another. There's an odd interaction between vtables and MMD for math operations. 20:38
I've put that aside briefly though because TT #389 really needs fixed.
How about other deprecations? 20:39
dukeleto chromatic: if you can send some examples to -dev about the bug with HLL subclassing and MMD, i can attempt to write more tests, if needed
whiteknight yeah, more info about the bug will put more eyes on it
chromatic Will do. 20:40
Coke I have some "ops to dynops" in my todo list. will probably not get to them until after rm_cflags branch is done.
bacek Coke, ship it!
chromatic Is that deprecation work fairly simple work, Coke?
dukeleto Coke: i refactored the RNG ops to dynops, so I can help with that if needed 20:41
Coke chromatic: pretty much, yes. anyone can chip in, just note on the ticket which ones you're taking, or ping me.
chromatic Okay.
20:41 allison joined
Coke allison! 20:41
chromatic Let's set priorities for the coming week. The next release is in two weeks. Which tasks and branches must we land before then?
allison Coke!
20:41 kurahaupo joined
whiteknight I think we can get the ICC branch finished and merged by the release 20:42
at least, that's the hope
Coke I'd realllly like to see rm_cflags done. =-)
whiteknight ditto
chromatic I'd like to get TT #389 and the MMD bugs fixed. 20:43
Any hope of the PCC refactors?
Tene here now
allison chromatic: possiby 20:44
I'm on my way home, and could work on it this weekend
chromatic whiteknight already volunteered to pick your brain and dole out tasks to a few of us.
allison this round will be smallest possible set of changes that can merge
my brain is already picked and documented on the subject 20:45
Tene Oh, tt389 still isn't done? I mentally filed that under "fixed and over" somehow.
I'll add 389 to my tasklist for this week.
allison so, anyone can jump in, I just figured they weren't motivated
perhaps we can pick a day for co-hacking on it?
whiteknight I'm plenty motivated. Still feel like the task is daunting
allison say, saturday?
whiteknight Saturday is good for me 20:46
Tene chromatic: is there anything about 389 that won't be revealed by 'make test' in the branch, or building rakudo against the branch or whatever?
chromatic It's not motivation on my part. I can't figure out what to do from your task list.
allison whiteknight: don't think of every possible task, just the smallest possible set for progress
chromatic: okay, is it lack of background on the tasks?
dukeleto allison: i think there is a disconnect on PCC stuff. i think a hackathon to get it done would be really useful
q1q 20:47
whiteknight q1q
Tene saturday is good for me, I think.
allison as in, would it help if I sat on IRC so people can ask questions, get specific tasks, and get followup on what's not clear?
chromatic I don't know if it's a lack of background. More like a lack of concrete direction to "How do I know if I've finished that one?"
whiteknight very helpful
Tene Doed the branch have the order of the ops switched in imcc yet? 20:48
allison sounds like a mini-hackathon is in order
I'll be around wed-friday too for anyone who wants an early start 20:49
chromatic Any objections to making that hackathon our primary weekly objective?
allison seems reasonable
whiteknight +1
mikehh +1
allison Tene: the branch has no changes yet, probably needs to be scrapped and rebranched
chromatic That's good, because it's one of two big things that Rakudo * needs for April.
Fixing TT #389 is another, because it's an annoyance they have to work around. 20:50
allison (my travel schedule has been bad since PyCon)
Tene what's the other?
chromatic The other big thing Rakudo * needs is a better GC.
20:50 bluescreen joined
chromatic If we can get to the sweepless GC after the PCC refactors.... 20:50
bacek Speaking of which...
dukeleto +1 to a hackathon weekly objective 20:51
Tene my expectation for the progress of the branch is that we first break everything by reordering the ops, and then we pick up the pieces.
bacek Sweepless GC will require additional housekeeping for VTABLE_destroy handling.
chromatic I don't think it will, bacek but we can talk about that later.
Objections to TT #389 as the other priority for the week? 20:52
Tene chromatic: none.
q1q
chromatic Let's move on to questions. 20:53
dukeleto?
dukeleto chromatic: I would like to volunteer to try to organize hackathons more regularly, and at some IRL conferences 20:54
chromatic You just did.
Coke yah, no one is going to vote against that.
mikehh excellent 20:55
dukeleto chromatic: I am organizing the 24hr Hacker Lounge at Open Source Bridge, and we really should have a Parrot Developer Hackathon there. It is going to be in an awesome space
bacek What about updating trac homepage to reflect "Weekly status"?
dukeleto chromatic: and since I just found out I am going to PGCON2010, that is another good place for one
Coke bacek: having to do timed updates on trac is going to fail.
(and already has.)
dukeleto crowns himself as Czar of Hackathons 20:56
Coke I would rather see folks post stuff on the news feed on www.parrot.org
dukeleto Is there any interest in having the @parrotvm twitter/identi.ca account updates shown on parrot.org ? 20:57
chromatic +1 20:59
allison dukeleto: as a sidebar?
bacek +0.05
allison twitter bits tend to be too meaningless for main features 21:00
dukeleto allison: yes, something unobtrusive
cotto_work +meh
allison dukeleto: sure
dukeleto allison: ok, i will make a TT
chromatic whiteknight, your question? 21:01
whiteknight Sent an email about this to the list: Would like to add a new vtable "find_class_method" that searches for a method on the Class or other metaobject, unlike find_method which searches for the method on the invocant.
we could improve a lot of encapsulation and add cool new features with this
fix MRO, allow new types of objects in MROs, etc
chromatic -1 from me, see my message to the list 21:02
whiteknight I'll reply to that on the list
other feedback, here or on list will be appreciated. EOQ 21:04
chromatic Tene, question?
Tene I haven't seen any response to my mail to the list. Is the answer just "Yes, subclasses of PMCs don't work right, ignore it and work around it for now."?
NotFound q1q
allison Tene: some subclasses of PMCs work, some don't. some subclassing problems can be fixed easily, some require more work 21:05
Tene Okay. 21:06
allison Tene: you may be running into the limitation that PIR subclasses can only access int, num, str, or PMC attributes of C PMCs 21:07
would have to look at it further
Tene nods.
allison but in general, we're trying to fix these problems with subclassing as we run into them
Tene I posted example code to the list, and a description of the problem. Depending on how I access it, an ATTR set on a subclass instance goes to one of two different places. Two sets of attrs, looks like. 21:08
allison it may be direct access to attributes somewhere instead of using the ATTR macros
especially, some of the code outside the exception PMC that accesses the exception PMC 21:09
Tene It does use the ATTR macros.
This is with just getting and setting attributes. The test case doesn't throw, doesn't use handlers, just instantiates and sets and gets attrs.
allison ah, more fundamental bug then
will leave it open on my browser and take a look tomorrow 21:10
Tene Exception.pmc has a 1-line get_pmc_keyed that dispatches to get_attr_str. using getattribute uses one set of attrs, using keyed which calls get_attr uses the other.
allison drops out to hop on a plane
Tene Okay. I'm done, then.
Have a fun flight.
Coke twitter block added to parrot.org 21:11
chromatic NotFound, your question?
NotFound nopaste.snit.ch/19839
chromatic +1 21:12
cotto_work +1
NotFound The other day some people said that will be good as an option, but that will be a lot more work. 21:13
chromatic What rationale is there for making it an option?
NotFound Don't know.
chromatic I can't imagine a downside.
Other comments? 21:14
21:14 cotto_w0rk joined
NotFound Well, I suppose we can tentatively commit it and revert if some problem appears. 21:14
Coke +1
NotFound Then I'll open a ticket about it and later commit, EOQ 21:15
chromatic Other questions? 21:19
Okay, let's call this meeting over. Thanks everyone. 21:20
bacek No other blocker/wishes from Rakudo*?
chromatic Those were the big December goals. 21:21
dukeleto goes back into #parrot
Coke ~~
21:21 dukeleto left
Coke drupal-- 21:23
ww
21:23 bacek left 21:25 chromatic left 21:26 NotFound left 21:38 cotto_work joined 21:53 bubaflub left 21:58 PacoLinux left 23:28 Whiteknight joined 23:51 cotto_mibbit 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!