Priority for this week: GC branches(test and merge), Google Code-In | Don't forget to post closed tickets in your report. | Note: This channel is only for our Tuesday status meetings (at 20:30 UTC); you probably want #parrot instead. | irclog: irc.pugscode.org/
Set by moderator on 16 November 2010.
00:45 bluescreen left 00:46 bluescreen joined, bluescreen left 00:47 bluescreen joined, bluescreen left, bluescreen joined 00:49 bluescreen left, bluescreen joined 02:09 whiteknight left 03:42 bluescreen left 07:07 bacek_ joined 07:12 bacek left 07:15 bacek_ is now known as bacek 08:41 bacek left, bacek joined 08:51 contingencyplan left 09:30 bacek left 10:02 bacek joined 10:49 bacek left 11:02 bacek joined 12:42 bacek left 13:03 kid51 joined
kid51 kid51's report 13:03
* Mostly worked on older tickets.
* Closed 3 tickets; took a couple of new ones.
* Discussed upcoming Parrot Developer Summit with cotto and whiteknight 13:04
* Will continue to work on Trac tickets I have taken 13:05
* EOR
13:14 kid51 left 13:30 bluescreen joined 13:51 contingencyplan joined 14:29 bluescreen left, bluescreen joined 14:33 bluescreen left, bluescreen joined 14:40 bluescreen left, bluescreen joined 14:42 bluescreen left, bluescreen joined 14:44 bluescreen left 15:15 bluescreen joined 15:38 PerlJam left, PerlJam joined 15:39 bluescreen left 15:45 bluescreen joined 15:49 bluescreen left 15:52 bluescreen joined 15:55 bluescreen left 15:58 bluescreen joined 16:00 bluescreen left 16:07 bluescreen joined 16:12 bluescreen left 16:17 whiteknight joined
whiteknight WHAT I DID 16:20
* Triaged a bunch of tickets and closed several (#1151, #993, #909, #827, #467, #1860, #581, #628, #642, #194, #299, #300, #1770, #548, #1450)
* Assigned myself to tickets dealing with HLLs, embedding/extending, and our users 16:21
* Stopped hashes from autovivifying (#1561). Will work on arrays next to try and get that ticket closed.
* Cleaned up some metadata in Trac
* Worked more on the embed_api branch. With my most recent commits (not pushed) We can build and run parrot normally again
* created dynlexpad_iter branch to make the DynLexPad PMC type iterable. Prototype code is in place, as soon as somebody writes a use-case or test for me I will be able to merge it (#1028)
* Created the get_user_id branch to solve issues in #731. Can merge and close if somebody can write me a test for the behavior
* Creating several dozen GCI tasks for enthusiastic students
* Started contacting Parrot users as part of the Product Management role to start getting feedback. Am thinking about holding semi-regular meetings for the purpose
* Sent out an email to the list asking people to focus on deprecations and experimentals for the 2.11 release. Several tickets I mentioned have already been closed. I have several questions to bring up today and in the next few #ps meetings
* Working to prepare for PDS on December 5th.
WHAT I WILL DO
* Continue working on the embed_api branch. Would like to get it usable, if not merge-ready by the end of the week. Will likely need some feedback and maybe a deprecation or two before final merge.
* Continue working on GCI, creating new tasks and mentoring students
* Continue working on ticket triage. Hoping to get down to 550 tickets or less (close about 50) before 2.11. Help is always appreciated.
WHAT I AM BLOCKING ON
* only 24 hours in the day.
16:41 mikehh joined 17:16 NotFound joined
NotFound What I did: 17:20
-parrot
* Implemented Parrot_gepid for win32 platform.
* Change set/get _addr to _label in test files.
* Added some tests.
* Closed some minor tickets.
-winxed
* Minor refactors, improving a bit the generated code.
What I will do:
No plan.
EOR
18:28 atrodo joined 18:53 bacek joined
Util # Done: * pbc_to_exe work, both NQP version and brain-dump. 18:58
# Plan to do: * Push pbc_to_exe branch to GitHub * Publish brain-dump on Parrot wiki * More pbc_to_exe improvements
# Blockers: * None
.end
(retrying)
# Done:
* pbc_to_exe work, both NQP version and brain-dump.
# Plan to do:
* Push pbc_to_exe branch to GitHub
* Publish brain-dump on Parrot wiki
* More pbc_to_exe improvements
# Blockers:
* None
.end
19:31 tcurtis joined 19:34 darbelo joined
bacek I merged gc_ms2_faster_list branch which make rakudo about 30% faster. 19:37
tcurtis I won't be able to attend #ps (school obligations interfere) today. I did nothing, and this weekend I'm hoping to spend some of my free time on getting my GSOC project back into the gsoc_past_optimization branch (and maybe even work on fixing any bugs if I have time). 19:40
19:47 nwellnhof joined
nwellnhof what i did 19:48
- had a look at the IO code, especially wrt Unicode
plans
- merge string_checks branch
- work on IO and Unicode
eor
19:54 plobsing joined
plobsing What I Did: 19:56
* deprecations (old ops, pir syntax, codestring)
* eliminate pbc interp constant hack
What I Plan:
* pmc backreferences in pbc constants
EOR
mikehh What I did since my last report: 20:12
* building and testing parrot on amd64/i386, with gcc/g++
* some fixes
* get version 2.10.1 fixes incorporated to get rakudo to build on latest parrot
* some branch testing and fixing
What I intend to do in the next week:
* testing and fixing
.eor
20:16 allison joined 20:20 kid51 joined
whiteknight hello 20:31
darbelo hi
cotto hi
bacek aloha
Util Hello
plobsing o/
NotFound Hola 20:32
mikehh hi there
whiteknight Okay, let's get started. How did we do last week? 20:33
bacek 2 releases, few branches merged, some tickets closed. 20:34
whiteknight :) Thanks for the recap, bacek
bacek plobsing++ did a great job ripping a lot of deprecated code
whiteknight yes he did. plobsing++ for the CodeString PMC deprecation 20:35
Our goals for last week were vague, as I remember. Getting a few branches merged probably satisfies
Util bacek++ merged a GC branch, which was a goal for last week 20:36
GCI was another goal, and whiteknight++ made progress on that.
(see /topic)
cotto q2q 20:37
bacek (I have to run soon)
whiteknight okay. I think that covers the goals from last week. Does anybody have questions (I have several)
mikehh have we got out git release +bugfix release procedure nailed down properly - there were problems there?
plobsing q1q
whiteknight mikehh: yes, I think we have it pretty well
Anybody else have questions to queue? Or should we start going down the list? 20:38
Util Ticket report: 40 closed, 18 new, 1 reopened
whiteknight 40 closed tickets is very good. Good job everybody
bacek for this week I would like my proposal of GenGC reviewed (carefully); Packfile PMCs design revamped; past_optimization merged. 20:39
c u
whiteknight okay. We have one goal for the coming week.
Questions. Cotto?
cotto bacek wants to merge gsoc_past_optimizations to get tcurtis++'s work back into master. Any thoughts?
whiteknight Does it need eyes/testing? 20:40
cotto I think it's ready to merge.
whiteknight okay. +1 from me then
cotto same here
tcurtis It isn't.
whiteknight ....damnit
cotto tcurtis, dang
whiteknight tcurtis: what state is it in?
cotto: second question (while we wait for tcurtis)? 20:42
tcurtis It's not up-to-date with github.com/ekiru/tree-optimization and in that up-to-date code there's a bug when supplying the :combine option to Tree::Optimizer.run. Other than that, I don't remember any other bugs. I'm planning to work on getting the parrot branch up-to-date this weekend so hopefully bugfixing and testing can occur. 20:43
cotto even if it's not perfect, getting it into master will give it more eyeballs
whiteknight maybe as a branch in parrot/parrot
cotto there's already a branch for gsoc_past_optimization 20:44
whiteknight oh, right. can we update that with the latest code?
plobsing we have a culture of pulling everything and anything into core. why do we *need* this in core at the current time? can't bacek rely on it as an additional dependancy?
mikehh that seemed to stall as tcurtis moved onto git with tree-optimization
whiteknight I think several people want to use it, and that it could become a pretty central part of the standard compiler infrastructure 20:45
plobsing: but you're right, we have lots of stuff in core that could easily be ejected
cotto we should consider coming up with a policy of what belongs in core and what doesn't 20:46
tcurtis has to go now.
Bye, folks.
whiteknight tcurtis++
plobsing several people wanting to use it now (but for some reason not being able to do so) feels more like a failing of plumage than an argument to pull it into core
whiteknight cotto: good point. Anybody interested in drafting such a policy?
kid51 What exactly does "pulling it into core" mean? 20:47
whiteknight kid51: including new things in the parrot repository
cotto kid51, putting it into runtime/parrot/library
whiteknight as opposed to keeping it in a separate repository
kid51 Those are two different answers
mikehh we already have a problem with core as certain ops are now generated by nqp-rx etc and that is NOT part of core
kid51 Something can be in runtime/parrot/library and be in master
20:48 tcurtis left
whiteknight nqp-rx is developed externally and snapshots are included in the parrot repo. 20:48
there are different degrees of inclusion
cotto whiteknight, I'll be thinking about what such a policy would look like.
whiteknight cotto++ 20:49
ok, cotto you had another question?
cotto Given that our long-term plan is to move away from a C-based PIR compiler to a self-hosted one (PIRATE), what do people think of removing pirc from the repo?
plobsing that was my question as well
whiteknight +1 from me. It's sad to see it go, but it isn't on the roadmap
NotFound +1 20:50
plobsing I'd like to see it extracted to a separate github project.
cotto I also hate to throw away work, but it's a sunk cost.
whiteknight plobsing: I agree, but I worry that it will just die there (as opposed to dieing where it is now)
I still think PIRC could be made to work without too much effort, but that's a gamble
cotto we can create a tag before ripping it out so it'll be easier to move to a separate project if someone wants to 20:51
whiteknight maybe we export it first, before just deleting it
plobsing I'd like to see it work, but I think the basic assumptions about it's role are the same as IMCC, which is most of the problem in the first place.
whiteknight does it's removal require a deprecation cycle? I don't think so since it's never really been used 20:52
cotto whiteknight, what do you mean by "export"?
mikehh it was supposed to replace IMCC but ran into a bus number problem
whiteknight cotto: move it to it's own repo, as opposed to just deleting it
cotto whiteknight, it's not part of the external interface afaik
wfm if we have a volunteer
whiteknight okay. no to deprecation then 20:53
plobsing me! me! pick me!
whiteknight ...plobsing?
I choose you!
plobsing I'm volunteering to extract
Util I would think PIRC would be a help as a reference to PIRATE implementers
whiteknight you done gots volunteered
cotto Util, PIRATE is largely working apart from the fact that it can't do op maps.
plobsing q1q 20:54
whiteknight okay, cotto your questions are answered?
cotto plobsing, can you post to parrot-dev once you've extracted pirc?
whiteknight, yes
whiteknight plobsing: question away 20:55
cotto I'm sure we'll have no trouble finding volunteers to rip it out.
plobsing can we deprecate PIR? by that I mean the assumption that a PIR compiler is always available in parrot?
whiteknight ah, I like that idea
and it may become necessary as the embedding API progresses 20:56
+1 from me
NotFound -1 for me while we don't have a replacement.
cotto plobsing, interesting idea. It needs some fleshing out, but I like where you're going.
mikehh agree with NotFound
plobsing you'd still be able to get it if you wanted, you'd just need to load it. you don't get it "for free" anymore 20:57
and by it I mean "compreg 'PIR'"
whiteknight plobsing: can you open an RFC ticket on trac to keep the conversation alive?
plobsing sure
whiteknight plobsing++
NotFound plobsing: fine with that
cotto plobsing++
whiteknight okay, any other questions before I ask all mine?
okay, lightning round 20:58
1) finalize opcode. (TT #1635) It's been experimental for several months. I suggest we make it official
NotFound +1
whiteknight it is the only defense we currently have against the inferior runloops problem. Entire swaths of Rakudo and other projects die without it 20:59
NotFound winxed has been using it all that time without any problem.
cotto +1 21:00
plobsing what happens to it when lorito hits?
whiteknight plobsing: at that point, all PIR ops are moot
NotFound plobsing: we can just redefine it as noop for a deprecation period.
whiteknight (but yes, the underlying inferior runloops problem goes away, theoretically)
cotto whiteknight, you might scare people with that attitude. ;) 21:01
whiteknight :)
okay, weak support. We'll discuss it elsewhere. Moving on 21:02
2) Parrot::Embed, I suggest we pull it out of the master repo and turn it into it's own project. parrot/Parrot-Embed on github maybe. I think it will get more exposure that way 21:03
plobsing +1
NotFound +1
cotto +1 21:04
NotFound But maybe we should wait until the embed/extend api matures a bit,
whiteknight I can agree with that
Util +1, if it returns to the main Parrot repo when it is done.
whiteknight Util: what do you mean?
like in ext/?
Util The extend APIs and embed APIs should live with the main code. Wait, am I confused as to what Parrot::Embed is? 21:05
whiteknight Parrot::Embed is the Perl5 module
to include parrot in Perl5
Util Doh!
whiteknight :)
Util +1
NotFound Parrot:Embed is for embedding parrot in Perl5 machine, isn'it?
cotto that's what the README says 21:06
whiteknight yes, it's a wonderful XS module
NotFound Sometimes we can't trust our READMEs ;)
whiteknight okay, that one passes. I'll ask one more question and save the rest for next week 21:07
3) find_codepoint PIR op. It's the replacement for some functionality originally found in Codestring. It's experimental, should probably be official. Vote to make it not experimental?
NotFound whiteknight: I think rakudo needs it.
whiteknight right
plobsing +1 # pge needs it
NotFound +1
21:07 bacek left
nwellnhof +1 21:08
but it requires icu, i think 21:09
whiteknight does it? We'll have to evaluate that then
cotto whiteknight, go on if you have more questions. We're only 40 minutes in.
whiteknight okay, I can do this all day
NotFound nwellnhof: a lot of things requires ICU and aren't experimental.
plobsing the op is still defined without ICU, but it throws unconditionally
code will still compile, but won't work if you try to use Unicode without ICU. seems reasonable to me 21:10
darbelo Most of our unicode support is a 'throwing stub' without ICU.
whiteknight 4) PARROT_LIBRARY and PARROT_INCLUDE env vars. They are marked experimental, but they've been in for a few releases now
do we continue to keep them supported, or do we not like them? 21:11
NotFound whiteknight: I think they can keep experimental.
plobsing they're crufty
whiteknight okay
darbelo Does anyone really use them?
NotFound darbelo: me, when expermienting. 21:12
darbelo Sounds experimental :)
whiteknight okay. We'll keep them as experimental for now, and bring it back up again later
5) OpenGL bindings. Do they belong in the master repo or should they be moved out to their own home? 21:13
NotFound People asked for "some way" to do something like that while command line fakecutables don't allow that functionality. There is a ticket about that.
The env vars, I mean.
mikehh there are problems there with long long
whiteknight NotFoundL if you can find that ticket, assign it to me 21:14
plobsing +1 to opengl being a separate project
NotFound whiteknight: I think is the one mentioned in DEPRECATED.pod
whiteknight Notfound: okay. I'll steal that one then
Util I will look into PARROT_LIBRARY and PARROT_INCLUDE working-or-not in fakecutables.
whiteknight Util++
mikehh there are quite a few bits that belong in plumage - BTW what's happening there
NotFound Util: no, the problem was not having the command line options that parrot main has. 21:15
The env vars are a workaround for that.
cotto what is the status of plumage? 21:16
whiteknight probably needs love
plobsing I have an alternative idea to solve the fakecutable arguments problem. I'll throw up an RFC
NotFound cotto: it worked fine last times I chekced.
whiteknight plobsing++ Great idea
Util Ah! I recommend shell scripts in place of the fakecutable for that situation.
whiteknight we probably need to get smolder reports working for plumage, then we can keep better track of it 21:17
cotto +1
mikehh 1
whiteknight any other opinions on OpenGL moving out, or do we shelve that till later?
NotFound cotto: I've been able to install winxed and blizkost from plumage and using Perl Gtk module from winxed. Not bad. 21:18
cotto NotFound, that's a good sign
whiteknight ...okay, I'll hold all the rest of my questions for next week. Let's move on to goals for this week 21:19
bacek wants people to read his new proposal for Generational GC. I suggest everybody read it, even if they don't think they will like it or understand it 21:20
plobsing where is it?
whiteknight It's on his branch somewhere. I'll post a link to the list
cotto I'll at least read that, if not try to give his code some review. 21:21
whiteknight The New York Times calls it "enthralling"
cotto github.com/parrot/parrot/blob/gene...c/gc_ms2.c <-this one? 21:22
whiteknight yes, that looks like it
any other goals this week besides gsoc_past_optimizations and generational_gc? 21:24
okeydokey then 21:25
mikehh GCI?
cotto packfile work
whiteknight ah yes, GCI is a good goal. We have a lot of tasks in there now, but we need a hell of a lot more 21:26
if we don't put out enough tasks, we won't be able to adequately take advantage of all that slave labor
cotto: is packfile stuff something we are going to do this week? 21:27
or are we just planning it this week? 21:28
NotFound I have a goal: people that asked for Parrot_getpid, please do some testing and report.
whiteknight :)
okay, I have to run now and it looks like we're losing steam. Call it a wrap? 21:29
NotFound +1
cotto yes
whiteknight gsoc_past_optimizations, generational_gc, GCI, and maybe packfile stuff are goals for this week.
mikehh anything on the summit? 21:30
whiteknight mikehh: We'll send out more emails (no responses to the last one)
mikehh seems to have lost track of the emails 21:31
whiteknight don't worry, there will be more
MANY MANY MORE
okay, I'm out. Good meeting everybody 21:32
21:32 whiteknight left 21:36 kid51 left, NotFound left 21:39 plobsing left
moderator Priorities for this week: gsoc_past_optimizations, generational_gc, packfile, Google Code-In | Don't forget to post closed tickets in your report. | Note: This channel is only for our Tuesday status meetings (at 20:30 UTC); you probably want #parrot instead. | irclog: irc.pugscode.org/ 21:42
21:49 atrodo left 21:52 darbelo left 23:30 nwellnhof 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!