"Tuesday at 20:30 UTC"
Set by moderator on 20 August 2010.
00:01 ascent left 00:13 ascent joined 00:42 tcurtis joined 01:29 bluescreen joined 01:45 whiteknight left 03:23 bluescreen left 03:25 bluescreen joined 06:14 cotto joined 07:12 tcurtis left 11:32 contingencyplan left 11:56 mariano__ joined 15:02 kid51 joined
kid51 kid51's report 15:03
15:03 ash_ joined
kid51 * Began service as member of board of directors 15:03
* (dukeleto, particle: need feedback on emails sent in last two days)
* Played role in closing several tickets, including about 4 of which I was owner
* Won't be available for #ps today due to medical appointment; will read backscroll 15:04
EOR
15:25 mariano__ left 15:49 mikehh joined 16:46 tcurtis joined 16:49 cotto_work joined 17:19 contingencyplan joined 17:41 ash_ left, kid51 left, spinclad left, tcurtis left, mikehh left, cotto left, Util left, TimToady left, Tene left, integral left, eternaleye left, wagle left, dafrito left, particle left, bluescreen left, ascent left, pmichaud left, PerlJam left, sorear left 17:43 davidfetter joined 17:45 davidfetter left, sorear joined, PerlJam joined, pmichaud joined, ascent joined, bluescreen joined, spinclad joined, kid51 joined, ash_ joined, tcurtis joined, mikehh joined, cotto joined, Util joined, integral joined, eternaleye joined, wagle joined, dafrito joined, particle joined, TimToady joined, Tene joined 18:01 kid51 left 18:10 NotFound joined
NotFound What I did: 19:27
-parrot
* Fixed a few bugs, some previously created by me
* Minor refactors
* Rearranged some tests with C coverage in order to give 100% coverage
when pass without compromising the tested features
* Added more core PMCs tests
-winxed
* Allowed calls to functions called 'new' in stage 1
* Minor internal changes
* Experimented with usage of perl5 modules via blizkost. Been able to run
helloworld examples with Gtk2 windows and buttons. Looks promising.
What I will do:
No fixed plan, maybe playing with blizkost
EOR 19:28
19:40 smash joined
Coke some board-related handoffs (more like handwaving). minor trac work, mailing list updates. 19:47
EOR
19:51 chromatic joined
tcurtis What I did: 20:06
* Uploaded my GSoC tarball to Google Code.
* Ran Rakudo spectests through Valgrind.
- Average of ~132kb "definitely lost" per spectest.
- Still working on more useful analysis.
* Didn't really get much else done on Parrot.
What I will do:
* No plan.
Blockers:
* Preparing for moving to Chicago next Saturday for university.
EOR
mikehh What I did since my last report: 20:07
* building and testing parrot on amd64/i386, with gcc/g++
* some fixes
What I intend to do in the next week:
* testing and fixing
.eor
tcurtis I may be late for #ps, so I'll go ahead and mention my suggestion for this week's priority: close $n old tickets (for some definition of old). 20:08
20:11 bluescreen left 20:14 luben joined 20:19 ash_ left
Util # Done 20:19
* On RosettaCode, Added Perl6 solutions for "Anagrams", "Formatted numeric output", "Top rank per group". Improved "Remove duplicate elements", "Rakudo".
# Plan to do: 20:20
* Work to close the neglected tickets that I own.
# Blockers:
* RosettaCode is addictive
.end
20:24 atrodo joined
chromatic I've been profiling and trying to tune. I could use GC-related help. 20:25
20:26 Paul_the_Greek joined
cotto_work #done: 20:29
- git o'clock approaches!
- set up a test site for github trac plugin at mksig.org/trac/ (apologies for slowness; it's due to the host vm for mksig.org)
- sent pull request for github plugin to upstream maintainer
- put draft migrtation timeline on the wiki under GitMigration, collaborating with dukeleto++
#hope to:
20:29 dukeleto joined
cotto_work - refine the github migration, get everything in place for the move 20:30
#eor
q1q
20:30 nwellnhof joined
chromatic Helly everyone. 20:31
hello
mikehh hello
dukeleto What I did:
* Worked on parrot github mirror.
* Worked on git docs :
github.com/parrot/parrot/blob/leto%...nology.pod
* Fleshed out the migration plan on the wiki
What I will do:
* Continue to work on git docs and refine the migration plan
.EOR
tcurtis Hi.
dukeleto howdy
cotto_work hii
Paul_the_Greek Hello folks. 20:32
chromatic How'd we do last week?
Util Hi
Paul_the_Greek Closed almost 25 tickets, I think.
NotFound Hola
dukeleto closed 23 tickets, i believe. very close to the goal
chromatic 25 is good. Was that about right, or was it a stretch>
dukeleto chromatic: i think it was about right 20:33
mikehh a bit - but good to aim for
cotto_work it was also a holiday weekend in the us
dukeleto is a proponent of aiming high
cotto_work: good point
mikehh a lot of the easier tickets have gone
chromatic Should we do it again?
mikehh yes 20:34
dukeleto chromatic: i think that sounds good
cotto_work +1
Util Exact count: 30 closed: fixed, 1 closed: invalid, 14 new, 4 reopened
chromatic 25 about right? 20:35
dukeleto chromatic: +1 to trying to close 25 tickets again this week 20:36
mikehh I think that should be our goal until we are down to around 299 or so open
chromatic How are we doing on branch merges? 20:37
dukeleto i saw detect_llvm get merged
mikehh was working on html_cleanup but $work intervened
nwellnhof i'd like to merge charset_massacre 20:38
dukeleto We currently have 32 active svn branches.
chromatic What's nearest to a merge point?
dukeleto Paul_the_Greek: is sleeker_boolean mergable? 20:39
Paul_the_Greek I think it's ready.
tcurtis Would it require a deprecation cycle?
chromatic sleeker_boolean? I doubt it.
Paul_the_Greek I talked with chromatic about that and we agreed this is compatible. 20:40
It passes exactly the same tests.
cotto_work ship it
Paul_the_Greek I also added a benchmark. 20:41
chromatic Other branches?
Paul_the_Greek Okay, I'll merge it this week.
tcurtis Branch pmc_multi_support can probably be deleted. It looks to have been merged into trunk at r46785. 20:42
chromatic Do it.
dukeleto tcurtis++
nwellnhof does everybody think it's ok to merge charset_massacre? 20:43
chromatic If it passes tests and doesn't hurt Rakudo/Lua, yes.
Benchmark figures would be nice too.
nwellnhof charset_massacre shouldn't give big speedups. it's mostly shuffling code around. 20:44
chromatic It shouldn't slow things down though.
nwellnhof i can check that. what's the preferred benchmark? 20:45
chromatic Anything string heavy.
nwellnhof i have looked for string heavy benchmarks, but couldn't find many. 20:46
chromatic Building Rakudo?
bootstrap-ops?
nwellnhof if building rakudo is a good enough benchmark, i'll go with that. 20:47
chromatic Any other branches? 20:48
make_html?
mikehh the indexing needs setting up, otherwise ok 20:49
20:50 allison joined
chromatic Any volunteers to fix that? 20:50
mikehh i.e. the index and sub-index pages from .json stuff already there
Coke s/building rakudo/passing rakudo's make spectest/ 20:51
mikehh I'll bring it up-to-date later and could use some help there -> html_cleanup branch
Coke note that we may have that many branches, but "active" is a strong word. see tools/*/branch_status.pl for a better idea of if they're active or not. 20:52
dukeleto Coke: only meant "active" in the "exists" sense 20:53
Coke: github.com/parrot/parrot/branches seems easier
Coke: it tells you last-modified date and how many commits behind/ahead of trunk (which is master in git)
chromatic Any other suggestions for next week? 20:54
dukeleto chromatic: Refine our plan to convert to git 20:55
If everyone can take a look at trac.parrot.org/parrot/wiki/GitMigration and see if we missed something, that would be great.
chromatic Oh, one thing we forgot to discuss. Memory leaks? 20:56
cotto_work some were found, probably not all of them 20:57
tcurtis I found a lot of leaking in Rakudo's spectests. I haven't tracked down the specific sources yet. moritz thinks that Rakudo skips global destruction, possibly explaining the leakage, but chromatic seemed doubtful of that. 20:58
chromatic Run VG on ./perl6 -e -1. If Rakudo skips global destruction, you'll see lots of leaks. 20:59
I don't see lots of leaks there.
21:02 dafrito left
chromatic Do we have tickets for those leaks, or is that in progress? 21:02
nwellnhof I ran most of the PIR files of the parrot test suite with valgrind, and there are many IMCC related leaks. 21:03
mikehh we seriously need to replace IMCC
Paul_the_Greek Should we perk up PIRC? 21:04
Coke mikehh: there are 2 well defined paths for that happen, and a 3 nebulous one. go for it. ;)
21:04 dafrito joined
dukeleto nwellnhof: it would be nice if you could email parrot-dev about some of the worst leaks 21:04
Coke dukeleto: as opposed to opening tickets? 21:05
nwellnhof it's mostly stuff like that: trac.parrot.org/parrot/ticket/748
always in the IMCC lexer
dukeleto Coke: both, preferrably 21:06
tcurtis chromatic: valgrind --leak-check=full does produce lots of leaks on ./perl6 -e -1 at rakudo g2ac5feb and parrot r48628, actually.
nwellnhof tcurtis: can you post the valgrind output somewhere?
tcurtis nwellnhof: for the whole spectest run? or for ./perl6 -e -1? The former(even after trimming to just the backtraces) is about 19k lines. 21:07
chromatic This is getting into specific development details that sound like #parrot topics.
tcurtis moves to #parrot. 21:08
chromatic Anything else on this topic for #ps?
cotto_work 1q
mikehh mee too 21:09
i.e. q2q
chromatic cotto_work?
cotto_work I'd like to propose that when we move to Git, we put our primary official repository on GitHub. dukeleto++ and I have been working on ironing out the details (GitMigration on the wiki), but I want to ask if there are any serious objections or concerns regarding our impending move to git.
I'll also send a message to parrot-dev asking for any objections/conerns later this week.
chromatic Why Github over OSU? 21:10
dukeleto chromatic: i see you haven't read #parrot from this morning ;) 21:11
cotto_work less work for osu
dukeleto chromatic: it takes us askign OSU for lots of things
chromatic: also, that can happen anytime we want, after the switch
changing our canonical repo is trivial in git, unlike in svn. 21:12
i have no problem mirroring github to git.parrot.org, but that can be done after the switch.
every clone is a full backup, so the problem isn't anywhere near as important as it was in svn 21:13
chromatic Other thoughts?
Paul_the_Greek Trivial question? 21:14
cotto_work anyone with a commit who's not in the Parrot org on github should talk to dukeleto about getting there.
chromatic Let's move on to the next question then. Paul_the_Greek, if it's about Git go ahead.
21:15 whiteknight joined
Paul_the_Greek No, it's about something else. I'll wait. 21:15
dukeleto has added just about everyone with a commit bit to github already. please let me know if you create a github account or if I missed you
chromatic mikehh, you had two questions. 21:16
mikehh I have been running make corevm/make coretest as part of my testing procedure for a while now
since the move of a lot of ops out of core was wondering if the utility of these tests
are as important now as they used to be - getting 2 tests fa]iling at the moment
our structuyre has changed quite a bit 21:17
structure
chromatic It's nice to have something simple when making deep changes, but if we're not making sure everything passes that's a problem. 21:18
mikehh I don't know if many other devs are running these tests 21:20
dukeleto I think the coretest/corevm targets are still very useful.
Paul_the_Greek Is github.com/parrot the place?
dukeleto mikehh: they are most useful when big changes are happening.
mikehh: they were indispensible when the calling convention refactor happened
mikehh also I was passing the tests on amd64 and founnd that they were pulling the libs from an older installed parrot 21:21
need to check on that
NotFound The failing tests are fixable by using the config hash from the interpreter instead of sysinfo, I think 21:22
mikehh I had rm'ed /usr/local/lib/libparrot.so but didn't remove other stuff 21:23
NotFound: I think so yes
chromatic Is there a ticket for that?
or a commit?
mikehh not yet
I was more worried why it was failing on i386 and not amd64 21:24
chromatic Can someone file a ticket for that then?
mikehh the library stuff or the test failures? 21:25
chromatic Either/both/whichever.
mikehh 'k will do
chromatic Next question? 21:26
mikehh dukeleto: that was when I included them in my test routine - after calling conventions refactor
mainly on smolder tests 21:27
I was wondering if anything had happened there
I thinkj we need them to look at tests on other platforms
we had quite a few problems with smolder.plusthree.com and it is now down 21:29
we need to run it from parrot.org 21:30
cotto_work it's installed now
mikehh we then need to set pointers to it
cotto_work I don't know if it's active. particle was the one who coordinated 21:31
mikehh and what about other things that used it - i.e. rakudo and others
21:32 kid51 joined
mikehh we need to get it up and running and encourage others to use it, especially other platforms 21:33
chromatic Other questions? Paul_the_Greek had one. 21:34
kid51 is home from medical appt
mikehh hi kid51
Paul_the_Greek Simple thing: Do we all agree that sprintf should format exponent with 2 digits? E+00
(unless more are needed, of course)
chromatic +1 21:37
mikehh what does it do now?
Paul_the_Greek It does that. It has been suggested that no leading zeroes be kept, so E+0.
But two digits is by far the standard. 21:38
This comes up because some sprintf's generate three digits, but we're trying to be consistent.
Two digits passes all the tests, one digit fails a bunch of them. 21:39
mikehh double and long double can give you more, but unless 3 are needed 2 is better 21:40
Util +0 # keep current behavior, until someone articulates why it is a problem on their platform.
kid51 Util: this is a long standing TT, so *someone* thinks its a problem
Util What is the ticket number? 21:41
mikehh I think 2 is the standard
Paul_the_Greek Hang on ...
trac.parrot.org/parrot/ticket/1200
kid51 TT #1200: opened by Coke 21:42
Paul_the_Greek I've added a #define for the standard number of exponent digits, so it's easy to change later.
21:42 ash_ joined
kid51 Paul_the_Greek: With an earlier patch from you I was getting test failures. Assuming we can get past them, +1 from me to standardization. 21:43
mikehh +1
Paul_the_Greek kid51: You mean an earlier patch on this ticket?
Util I remember working on that - platform-independent is the way to go. +1 to 2 digits (unless 3 are needed). 21:44
21:45 ash_ left
Paul_the_Greek The code now canonicalizes to 2 digits unless more are needed. It works for any number of exponent digits. 21:45
chromatic Any further questions? 21:46
kid51 It seems that Util and Paul_the_Greek have different positions. I doubt we're going to settle it in today's psketch.
Util: If your position differs, please post in TT #1200
Paul_the_Greek I think we're agreeing.
kid51 k 21:47
Paul_the_Greek 2 digits unless more are needed.
Util I agree that we are agreeing.
kid51 k
chromatic Anything else? 21:48
21:49 plobsing_ joined
chromatic Let's wrap it up. Good work everyone. 21:50
21:50 luben left
dukeleto ENODISHES 21:51
21:51 dukeleto left, Paul_the_Greek left 21:53 atrodo left 21:56 NotFound left 21:57 kid51 left 22:12 ash_ joined 22:15 plobsing_ left 22:30 tcurtis left, cotto_work left 23:21 bluescreen 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!