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 1 February 2014.
12:12 denisboyun joined 14:41 bluescreen joined 15:30 isBEKaml joined 18:13 denisboyun joined 18:18 denis_boyun_ joined 18:57 sa1 joined
Util # Done: 19:27
* Taught Blue_Tiger how to:
+ Change `foreach` into arrow form
+ Change mapish EXPR to { BLOCK }
# Plan to do: 19:28
* Presenting the Perl [5|6] talk (that I gave at Perl Oasis) at atlanta.pm.org/
* Participate in GSoC planning:
+ github.com/perl6/mu/blob/master/mi...4/ideas.md
+ wiki.enlightenedperl.org/google_sum..._code_2014
+ groups.google.com/forum/#!forum/tpf-gsoc
.end
19:29 PerlJam joined 19:30 rurban joined
rurban hi 19:30
Util Hello! Who all is in attendance?
Hi, rurban 19:31
sa1 Util: hello
Util Hi, sa1. Any plans to participate in GSoC again this year?
sa1 was/is still stuck hunting for an internship, and was unable to proceed on those bugs.
Util I understand. 19:32
sa1 Util: I can help out but there's a compulsory internship and I can't divide time.
I don't see anything parrot-specific in those links.
Util sa1: I take that to mean that your compulsory internship will prevent you from being a GSoC student at all this year, correct? 19:33
sa1 yeah
Util sa1: Those links are just being put together in the last few days. Parrot-specific entries have not been made yet.
sa1 okay 19:34
Util I suspect that (at least partially) the Parrot and Perl groups will be running a joint GSoC effort, like last year.
rurban See wiki.enlightenedperl.org/gsoc2014/ideas Re parrot: I applied a configure patch #1030 and plan to fix the perldoc root issue soon.
Util Not definite yet
sa1: perhaps you might have time to help as a non-student this year. When the new students get to the part about makeing 19:35
sa1 Util: yeah, sure. 19:36
Util a detailed inchstone plan, for example, they will value your input as a former participant
sa1: great!
rurban: thank you for fixing that libffi test, and TIA for the perldoc fix! 19:38
rurban I also worked shortly on the rurban/gh1017-proxy-get_id fix and testcase, but it still fails with threading 19:39
Util rurban: do you have any Parrot projects that you would like to propose and montor this year? I see you are already down for some P5&P6 items.
rurban I'm pretty busy still with perlcc for a week (closing the last p5 core testsuite issues) 19:40
Util ok
rurban parrot none, sorry. but we should mention smthing like a parrot speedup project 19:41
Util we have until the 14th to finialize our org in GSoC.
rurban unfortunate run-time signature handling and internal sprintf's, esp with unicode strings 19:42
Util rurban: What do you think is a manageable chunk that could be accomplished by a new student
rurban that's what drove moarvm away
Util i.e. what area to target in Parrot?
rurban hmm, sol should decide
ah, sal, sorry
Util Low hanging fruit, but not trivially low.
sa1 its sa1 :) 19:43
rurban and bring back in the optimizations coke removed. they were stable
Util what were the nature of those opts? and why did he remove them?
sa1 rurban: its not going to be my project so I guess the new student may be able to judge better. 19:44
rurban at least -O1 to -O2 were stable. -O3 had a bug which a fixed in branch after the removal
why? general crazyness, I guess. as typical in perl projects
Util I agree about "run-time signature handling"; it looks like a good payoff for the amount of work, and achievable.
rurban the signatures were good, but then it was refactored some years ago to get simplier and slower and slower. so it's a "reverting history" project 19:46
Util Not just reverting history; looking at the old way and new way with an eye for absolute best performance, since that area has been identified as a hotspot. 19:47
rurban or fixing NQPLexPad for parrot threads. short but complicated. 19:48
github.com/perl6/nqp/issues/67
this was also a nqp blocker for parrot threads 19:49
Util Sounds good. 19:50
rurban: You mentioned speed issues with internal sprintf. Is there an open ticket, or discussion that you could point me to? 19:51
rurban With pmichaud we analyzed and profiled parrot slowness in Norway 19:52
one was the de-optimized sigs (the nci way) 19:53
the other was our internal handling of arguments, which goes through several stacks of sprintf calls, at run-time.
and bad layout decisions/alterations in unicode encoding throughout imcc 19:54
something like this if I remember correctly, nothing written down. 19:55
maybe there's a email to the parrot list somewhere
Util OK; I had thought that you meant that the Parrot sprintf function was badly non-optimal. I understand now.
rurban but that's a too big project for a newbie GSOC student. on the side hugmeir from p5p could have done that 19:56
Util Do you think that just the "run-time signature handling" piece, without deliving into the sprintf parts, is managable for GSoC? 20:00
rurban yes, for a good student yes 20:01
because it's not much new work to do
I can act as mentor also 20:02
Util I want more out of the project than just a return to the old, if possible; however, I might be wrong about the possible speedups available over-and-above even the *old* code. 20:03
In any case, I am glad to hear you say that you can mentor.
We are past 30 minutes, and running out of steam on this thread. 20:06
Anything else to discuss before adjourning? Anyone?
sa1 not from me, but if there's any other work required during org application period, I'm a ping away. 20:07
Util sa1: much thanks! 20:08
OK, then. Meeting adjourned. Thanks to all! 20:09
23:32 lizmat 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!