Post closed tickets in your report. | Note: This channel is for our weekly status meetings (Fridays at 13:30 UTC); you probably want #parrot instead. | irclog: irclog.perlgeek.de/
Set by moderator on 21 January 2015.
06:40 rurban joined 08:46 rurban1 joined 12:29 rurban1 joined, Chirag joined
Util Pre-report: 13:27
# Done:
* Released Parrot 7.0.0
# Doing:
* Reading up on the 7.0.1 hotfix.
.end
rurban1 2015-01-23 rurban 13:31
DONE:
* merged smoke-me/rpa-splice-shrink-fast-gh1174, perl6 passes now all spectests
* releases 7.0.1 hotfix, with 4 fixes
* GC --optimize segv #1186
* threads deadlock #1187
* smoke-me/m64-gh1181 #1181, #1182, #1178, #1110
* Actually link with libicu when probing for it. #1182, #1188
* merged smoke-me/gcc_cmpxchg-gh1173 atomic now usable (for threads)
* merged Optimize away ExtUtils::Command on posix systems. #1177
* online doc updates
* added smoker and number scripts to parrot/misc
* updated timeline.md in parrot/misc, added a NCI task
* updated parrot-bench for 7.0 and stringbuilder
WORK:
* rurban/stringbuilder-gh1123, 10% faster in benchmarks
(no need to clone, maybe more substr optims possible)
* smoke-me/thread-lock-gh1187, minor refactor, but deadlocks
BLOCKED:
7.0.1 was needed for machines with low memory, like 2GB 13:32
that's why we didn't catch it
thinking about adding one gc stress test to the normal suite 13:33
this would have caught it.
I still got one more GC bug, on my small centos5 system.
Util Hello! Who all is in attendance? 13:35
rurban1 Me
Chirag Me (lurking) 13:36
Util Welcome! 13:37
rurban1 7.0.1 is now the new default in rakudo. much faster than the old parrot, all spectests passed
lizmat bumped rakudo to 7.0.1
rurban1 thanks!
lizmat is off to lunch&
Util rurban1: congrats on getting 7.0.1 fixed in time for the Rakudo release. 13:38
rurban1 yep, close
Util lizmat: thank you
rurban1 I held it back for one full day to wait for smoke results
caught some more crazy configure test errors 13:39
Util rurban1: +1 to "adding one gc stress test to the normal suite".
rurban1 yes, probably the first. it needs to be a good and stable one.
I'll make a ticket for it
#1189 13:42
next work will probably be threads. I'm not so sure if we can go away without atomic updates of our primitive values (int, number) 13:43
the scheduler locks, we could get rid of some locks I believe with atomic updates. if it's faster. atomic is pretty costly 13:44
but in practice our threads are still pretty good. better than moar and jvm for sure 13:46
but the jvm has a much better garbage collector and faster ops 13:47
Util Reducing locking... Hmmm... Do you have any true multi-processor (as opposed to multi-core) machines in your smokers? 13:49
I would think that multi-processor testing would be required when reducing locking, but I could be wrong about that. 13:52
rurban1 no, multi-core is the only useful testcase we need 13:53
Util Great
rurban1 99% intel is our usecase. maybe ppc64 8 or armv8 later, when they will be used in the cloud
I'm looking into threads mainly for my potion vm, libuv got now nice threadpools, like parrot 13:55
Util Two minor notes about 7.1.0's date: 13:57
* The RMGuide originally said "Feb 16, 2015 - 7.1.0 - rurban"; I changed it to "Feb 17" in f9bb25b to match the Tuesday normal schedule.
* The hotfix release said "Our next scheduled release is at 23 Feb 2015."
rurban1 oops, 17. it is 13:58
I did it right in the github milestone only
thanks for catching this. but too late 13:59
Util No harm done; I think the only group that cares about our "next scheduled release" date is Rakudo, and they won't be changing their release schedule next month.
Anything else to discuss? From anyone? 14:02
OK. Thanks everyone! 14:05
Meeting adjourned.
14:36 rurban2 joined 16:23 rurban joined 16:49 rurban_ joined 21:36 rurban 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!