Perl 6 - the future is here, just unevenly distributed

IRC log for #parrot, 2014-08-04

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:45 FROGGS_ joined #parrot
01:51 Chirag joined #parrot
01:53 Chirag Coke: Hey! the build failure is on my Dev branch only (pcc-gh1083)
06:35 bighugedog joined #parrot
07:32 basiliscos joined #parrot
08:37 bighugedog joined #parrot
10:22 bighugedog joined #parrot
11:05 bighugedog joined #parrot
11:14 kid51 joined #parrot
12:15 atrodo joined #parrot
12:37 FROGGS rurban: when I see suh a thing in a bt for a perl6-p program:
12:37 FROGGS Parrot_ext_call (interp=interp@entry=0x60f050, sub_pmc=0x35dce88, signature=signature@entry=0x7ffff7b2e18c "->") at src/extend.c:158
12:37 FROGGS that looks like it is something about rakudo's ops, or maybe even nqp...
12:38 FROGGS how can I find out what is called there?
12:40 FROGGS the problem I try to hunt down is that we have something of type enum_class_Sub, which has PMC_data==NULL
12:41 FROGGS the behaviour is identical on parrot 6.1.0, 6.4.0 and 6.6.0... now switching to HEAD...
12:52 FROGGS the entire backtrace is here: https://gist.github.com/FROGGS/6a3d23f0b93db0a9938b
12:53 FROGGS this issue causes more than 1600 spectest failures btw
13:06 FROGGS my gut feeling tells me that this might be related: https://github.com/rakudo/rakudo/blob/nom/src/vm/parrot/ops/perl6.ops#L1049
13:51 Chirag joined #parrot
14:47 rurban1 joined #parrot
14:49 rurban FROGGS: lemme check...
14:50 rurban Parrot_ext_call is a normal method call (not vtable). The backtrace is not really helpful. Will try with a debugging build
14:54 rurban nqp and rakudo HEAD?
15:10 Chirag rurban: Hey! I am getting a build failure with my branch -
15:10 Chirag too few positional arguments: 2 passed, 3 (or more) expected
15:11 rurban I'll check soon
15:11 Chirag ok
15:11 rurban first checking the perl6 problem
15:11 Chirag sure
15:57 rurban FROGGS: not repro. most spectests pass with nqp+rakudo HEAD and parrot 6.6.0 debugging. Messing with PERL6LIB?
16:02 rurban Chirag: can repro $ ./miniparrot -Iruntime/parrot/include config_lib.pir
16:02 rurban too few positional arguments: 2 passed, 3 (or more) expected
16:05 rurban This is in the very first call to fill_params
16:08 rurban with the sig PiSSoIpSnSoIpSnPoIp (Parrot_IMCCompiler_nci_compile_file)
16:08 FROGGS rurban: yes, nqp and rakudo HEAD will do
16:09 FROGGS no, PERL6LIB is not set here
16:09 FROGGS it also fails on other boxes
16:09 rurban interesting
16:10 FROGGS like Coke++'s smoke box: https://github.com/coke/perl6-roast-data/blob/master/log/rakudo.parrot_summary.out
16:14 Coke I don't have the specific failure modes, only that the tests abort there.
16:15 Coke i can get the specific failure modes, though. FROGGS - did you try reverting the rakudo commit I Id'd in my ticket?
16:15 Coke (we still shouldn't be segfaulting, but that might help narrow down what the issue is)
16:16 Coke https://rt.perl.org/Ticket/Display.html?id=122436 - the one example I bothered to pull from the spectests segfaults there.
16:17 FROGGS Coke: no, I've not tried that
16:17 FROGGS Coke: will do that now
16:18 Coke just posted the segfault again in #perl6; thanks for helping track this down.
16:25 FROGGS Coke: reverting it helps
16:38 FROGGS this is how it looks like when I invoke it several times: https://gist.github.com/FROGGS/5a8688b59c1b06137ec1
16:39 FROGGS (using Test.pm instead of Test.pm.pir)
16:39 FROGGS I'll strip down Test.pm now until it goes away
16:47 Chirag rurban: just went through that code but we never modified it.. (after the last task)
16:49 FROGGS rurban: hmmm, using a module triggers it (in combination with 'say :10("111")'), the module can be empty though
17:25 Chirag rurban: I suspect some problem with Parrot_CallContext_class_init
17:26 Chirag oops.. looks like you have your hands full :) .. will wait
17:51 Coke jnthn was also seeing a build failure ( I think ) on nqp-parrot on windows, which made it difficult for him to get near the segfault FROGGS is looking at; not sure if at a ticket was opened. I don't have a handy windows box to test.
18:15 FROGGS rurban: here is also now valgrind's output of a good and a bad run: https://gist.github.com/FROGGS/6a3d23f0b93db0a9938b
18:46 basiliscos joined #parrot
19:34 rurban Looks like a Parrot_pcc_reuse_continuation bug
19:36 zarchne joined #parrot
19:47 rurban very interesting, needs a ticket and a bisect probably
22:23 kid51 joined #parrot
22:31 Psyche^ joined #parrot

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced