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 18 July 2012.
13:13 bluescreen joined 14:42 benabik joined 16:34 jashwanth joined 18:20 lucian joined 18:41 rurban left 18:45 benabik joined 19:01 rurban joined 19:07 benabik joined 19:20 brrt joined
rurban who will take part? 19:30
cotto howdy 19:31
brrt hi 19:32
rurban so whiteknight is probably off power in a storm 19:33
cotto he has all the fun 19:34
rurban I'm not sure if we should wait. io_cleanup1 can be merged, threads probably not. 19:35
cotto excellent 19:36
rurban ok, I'll start. I have a big list.
DONE: 19:37
* threads: avoided sleep in nci tests (fixed #807), but still
ongoing races in sleep with threads and signals.
* threads: found a runloop_id_counter race with tsan #810, which nine fixed.
blogged about the new tool tsan.
* published qemu images to smoke powerpc, mips32, sparc32.
perl514.cpanel.net/qemu/
* io_cleanup1: reproduced and fixed t/pmc/filehandle_28 swapfs problem in #811.
smoked solaris.
* tested and merged lndir #805
TODO:
* nci generator produces bad prototypes (C++ compilers only) #806
* analyse and fix native_pbc endianness (untested feature) #807
* ppc neg_0 failure (obscure platform)
* I might volunteer for the september release
BLOCKS:
* get a (ppc/mips/sparc)64 qemu/HW for native_pbc tests, #394.
There is ongoing upstream qemu/openbios development.
* I'll be in Europe in extensive p5 meetings for the next 3 weeks.
* Looks like there will be a p5 "Austin compiler group" formed,
because this is the topic for YAPC::US 2013. This will take a lot
of my time.
EOR
oops, I forgot: I will doing a address-sanitizer talk at YAPC::EU 19:41
brrt my report 19:42
DONE:
- finalized the reorganisation of mod_parrot so we have a lot more new extension points
- blogged about it
TODO:
i guess i'm throttled
DONE
- finalized the reorganisation of mod_parrot so we have a lot more new extension points
- blogged about it
TODO:
- reorganise routing (i.e., figuring out what part of which script to call) (pretty much a thinking issue)
- add license information 19:43
- update documentation
EOR
rurban cotto? 19:45
cotto still eyeballs-deep in $dayjob
rurban Coke?
nine: I know you could not repro the remaining threads failures. I see them all over windows/cygwin. 19:46
Should we merge? should we leave it to whiteknight? 19:47
Coke rurban: yes?
rurban Coke: Anything to report? 19:48
Coke only thing that is remotely parrot related on my plate right now is the coke/rm_pasm branch, which is already outdated and needs a good merging.
if there is a desire to move forward on that, I can try to bump it up on my todo list. 19:49
I started it to see how far we could get with ripping it out. (which, turns out, is "pretty far.")
rurban Coke: merges cleanly. I'm smoking it now 19:50
nine: threads? 19:51
Coke there have been new things written in master since last I touched it that required some pasm care and feeding, I think.
rurban++
rurban I would say: we need someone with signal experience to test the threads races I'm seeing. before we merge threads. 19:52
Coke: interpinfo.pir is missing 19:53
Coke yup, that's it. needs to be converted in the branch to PIR from PASM. 19:55
rurban in h2inc.pl? 19:56
threads: whiteknight said he is seing no race in the loop I pointed out from the backtrace. nine was saying, that he has no signal experience, he only knows threads, and threads look good to him. better than before. 19:59
Coke: yes, rm_pasm needs a little bit of love. September? 20:02
who else? benabik? 20:03
nobody else? so whiteknight will probably only merge io_cleanup1, and that's it. 20:06
brrt would like to have multipe interpreters fixed, but thats not super-realistic right now 20:08
rurban some disagreement anywhere?
brrt no, theres no certainty as to the cause 20:09
rurban Is there a branch or issue to fix multi_interp?
brrt because its basically hanging on finalization which seems to be working on a corrupt data structure
rurban Oh, I remember. gc is also an issue here.
brrt threads fixes the first half, find_root_scheduler the same half
but the finalisation thing is a mystery
rurban We have several tickets on this. 20:10
big red ones
rakudo definitely wants us to prioritize this 20:11
brrt so do i, but its 'hard to debug' 20:13
rurban we definitely need a 90% whiteknight
brrt and, not to be mean about it 20:14
i'm not sure anyone really knows how multiple / child interpreters are 'supposed' to work
rurban brrt: Could you mail the list about the issues? 20:15
brrt yeah, sure :-)
anyway, i'll try to get the appropriate tests into parrot 20:16
rurban ok, so let's call it a wrap, and hope whiteknight gets some power to do a release next week.
21:24 lucian joined 22:03 benabik joined 22:15 bluescreen joined 23:30 whiteknight joined