Parrot 4.6.0 "Wild Parrots of Telegraph Hill" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 18 July 2012.
01:04 woosley joined 01:06 kid51 joined 01:25 benabik joined 01:44 benabik joined 06:24 awwaiid joined 07:13 brrt joined
dalek kudo/nom: fefacce | moritz++ | src/core/IO.pm:
[IO::Path] fix handling of local paths in .new(Str)
 07:15
07:41 Timbus joined
brrt so, i still don't know why the gc bails on multiple interpreter construction 08:23
s[con]=des 08:24
08:30 Khisanth joined 10:30 brrt joined
dalek kudo/nom: 78c5c17 | moritz++ | / (2 files):
Detect and carp at circular module loading
 10:33
11:07 JimmyZ joined
dalek kudo/nom: 34e8d4d | moritz++ | src/Perl6/Actions.pm:
guard infix:<but> against zero-arg calls on the RHS

closes RT #73990		 11:49
12:13 mtk joined 12:37 sivoais joined
Coke brrt: as in "don't know the exact technical cause" or "how did this ever work?" 12:42
(because it probably never did, is my guess) 12:43
aloha?
seen aloha?
aloha aloha was last seen in #perl6 2 days 7 hours ago joining the channel.
Coke msg bacek to please disable the "explain" command in #perl6, as well as its reaction to negative karma. Thanks.
aloha OK. I'll deliver the message.
Coke explain yourself
aloha positive: nothing; negative: nothing; overall: 0.
brrt Coke: i"m backlogging to what i said 12:44
Coke: I don't really know why it doesn't work
assuming the GMS collector, that is 12:45
seems gc_sys->gc_private is initialized to the Wrong Thing in a child interpreter
Coke well, you have how many days until pencils down? If this is a sticking point that isn't going to get resolved, you probably have to route around the damage. 12:46
(come up with a modified plan that still gets something done by end of GSOC)
and then come back around to do it the right way later. No?
brrt soft pencil down = next monday, hard pencil down = monday after that
well
there is no real route arround is
afaik
arround it
Coke you're using multiple interpreters to do it the "right" way in mod_parrot, yes? 12:47
so do it the wrong way.
Having that, with the known caveat that we're doing it wrong and it will be fixed when parrot is fixed seems reasonable.
brrt its already done in the wrong way :-)
the point is
there are actually multiple points 12:48
one: the wrong way cannot run in a multi-threaded server
because i cannot create interpreters independently and destroy them independently
(this is the main point)
two: i cannot get a separate interpreter for a loader and a user script
(same reason, i cannot destroy the interpreter once made) 12:49
hence
i must run both (loader and script) in the same interpreter
Coke have you written a failing test for parrot's test suite that shows the issue?
that will help whoever can fix the problem, anyway.
brrt yes, but haven't integrated it into parrot test suite
its in a separate repository 12:50
whats slightly more, master parrot cannot run scripts in a child interpreter period
threads parrot can, though
and my own hacky fix can too
12:51 darbelo joined
brrt basically whoever can debug it will have to have some deep understanding 12:51
i can obviously do other things on mod_parrot
... but not such obvious ones :-)
reorganising the structure of it all so multiple interpreters can be added at a later time, that i can do 12:52
and just lived with short-lived single interpreters for now
because that does work
rurban threads on linux also broken? asan smolder.parrot.org/app/projects/rep...ails/30427 13:05
JimmyZ rurban: pass here 13:17
rurban With asan on threads?
mips32 / threads did also hang in task.t 13:18
Maybe we have to increase the sleep for old slow machines.
JimmyZ rurban: I just typed `perl Configure.pl --optimize && make -j4` 13:19
13:26 bluescreen joined
rurban I think it's related to the CPU speed. asan slows down runtime by 2. I'll try longer sleeps. 13:27
JimmyZ doesn't know what's asan 13:33
rurban code.google.com/p/address-sanitizer...sSanitizer 13:34
JimmyZ 'oh 13:35
rurban: I did'nt use asan 13:42
13:52 tobym joined 14:56 dmalcolm joined 15:23 whiteknight joined
whiteknight good morning, #parrot 15:33
brrt good afternoon whiteknight 15:36
whiteknight hello brrt. How are you doing? 15:37
brrt as far as my bugs go, same situation as yesterday :-) 15:39
but ive decided i can re-organise interpreter handling anyway
and just fall back to the single-threaded case
whiteknight yeah, sorry about that. I lost a lot of time last night dealing with some personal stuff 15:48
15:51 ligne joined
dalek p/toqast: 6849356 | jnthn++ | / (2 files):
Add QAST::VarWithFallback, a way to give a handful of variables a fallback way to do a lookup should the initial lookup return null. For use of things like contextuals, or when the fallback is to throw an error.
 15:52
p/toqast: 43dfb24 | jnthn++ | src/QAST/Compiler.nqp:
Compilation of QAST::VarWithFallback.
p/toqast: e714e6e | jnthn++ | src/QAST/Compiler.nqp:
Corrections to contextual scope compilation.
p/toqast: 64f922a | jnthn++ | src/NQPQ/Actions.pm:
Get a basic cut of contextuals working again. Fixes the contextuals tests entirely plus another test file that depended on them.
brrt oh don't worry about that 15:57
its hard
i know
:-) personal stuff happens to me all the time, too 15:58
(i'm off, hope to be on #parrotsketch tonight!) 16:01
whiteknight ah, thanks for the reminder
Coke whiteknight: does the GCC compile farm help you with your BSD needs? 16:05
whiteknight Coke: I don't currently have access to it. I'm told it's easy enough to get though 16:10
Coke ISTR dukeleto was the guy with the keys. (or the one who knew where the keys were, anyway.) 16:12
whiteknight right, and he's been busy 16:18
16:23 brrt joined 16:24 jashwanth joined
dalek p/toqast: 207256a | jnthn++ | src/QAST/ (2 files):
Add a (per-HLL configurable) to do positional and associative lookup/binding using a QAST::Var. This may mean that even Rakudo can use this approach for its array/hash accesses, by configuring them to call the appropriate methods. Doing things that way come in highly useful with analysis/optimization of native array accesses.
 16:29
p/toqast: 90adf5a | jnthn++ | src/NQPQ/Actions.pm:
Fix array and hash indexing; wins back some more tests.
p/toqast: 0143bc0 | jnthn++ | src/NQPQ/ (2 files):
Assorted fixes to package scoped things. Fixes our subs amongst other things.
 16:46
16:47 benabik joined
dalek rrot/native_pbc: a9dde2f | rurban++ | / (8 files):
[CAGE] fix most t/codingstd/copyright.t issues

Only imcc has some Copyright (C) 2002 Melvin Smith <melvin.smith@mindspring.com> left.		 16:51
rrot/native_pbc: 87e856d | jkeenan++ | / (8 files):
Move Parrot_cpu_type function out of src/platform/misc.c.

This way, it doesn't need to be repeated elsewhere. For symmetry, also remove it out of win32/misc.c and give it its own file there as well.
Patch supplied by Andy Dougherty++.
p/toqast: 50bb7ae | jnthn++ | src/NQPQ/Actions.pm:
Eliminate a few more viviself usages and remove vivitype. These updates fix a few more package scoping issues, winning a couple more test files back.
 16:55
17:18 ligne joined
dalek p/toqast: 3f525aa | jnthn++ | t/nqp/38-quotes.t:
Update a pir:: op in a test to the appropriate nqp:: one.
 17:23
p/toqast: eeeaea0 | jnthn++ | src/NQPQ/Actions.pm:
Fix array constructor.
p/toqast: 2524810 | jnthn++ | t/nqp/40-lists.t:
Update a test to not use pir:: ops.
p/toqast: 4660299 | jnthn++ | src/ (2 files):
Add an nqp::ishash.
rurban my smoker: gist.github.com/3287531 17:26
18:31 lucian joined
dalek p/toqast: 737fd97 | jnthn++ | / (4 files):
Get ++ and -- back in place; since we don't have a container model, it really maps down to binding and stuff. Dubious, but removing it will probably cause riots...
 18:44
18:44 benabik_ joined 18:48 benabik joined 18:49 tuxit joined
cotto #ps in 15 19:15
19:16 rich joined, rich left
dalek CT: f5b312d | benabik++ | src/PACT/Packfile/Decompile.winxed:
Packfile.Decompile: subs don't have get_bool

I wanted to check for null anyway		 19:25
CT: 47b2f25 | benabik++ | src/disasm.winxed:
disasm: Initial assembly output

Probably more intelligible than the dumper output.
benabik gist.github.com/3288622 19:26
whiteknight benabik++ 19:28
cotto benabik++ 19:30
dalek CT: fd43c3d | benabik++ | src/disasm.winxed:
disasm: output string encodings
 19:31
CT: 1e6843f | benabik++ | src/disasm.winxed:
disasm: minor change to null output
19:54 brrt joined
benabik okay, now parrot is hanging loading the decompiler. *sigh* 20:07
whiteknight wonderful 20:08
benabik Oh, wait... no the decompiler is hanging on something else.
Helps to advance the key instead of just doing the same part over and over again. 20:24
Yay handling multis... 20:25
Woo, decompiling non-trivial PBC 20:31
dalek CT: 096353c | benabik++ | src/PACT/Packfile.winxed:
Packfile.Key: Actually advance the key
 20:34
CT: f6ce9b1 | benabik++ | / (3 files):
Packfile.Decompile: Initial multi handling

Appears to collect multi candidates properly
CT: 98f4468 | benabik++ | src/PACT/Packfile.winxed:
Packfile.Key: use int for types

Fixes calling the wrong multi constructor
brrt i'm having abstraction fever
CT: 1062de4 | benabik++ | src/disasm.winxed:
disasm: No longer need dumper.pbc
CT: 05dc11a | benabik++ | src/disasm.winxed:
disasm: Print multi candidates

Oh, it does it wrong, but it at least shows them
benabik installable_disasm can now decompile Test/More.pbc
brrt i'm looking at all this code i've written and i'm thinking 'hey we could add a layer of indirection there'
benabik I think I need to replace the Subs in the constant table with Packfile.Subs 20:38
But first, I need to rest, my wrist hurts again. 20:43
gist.github.com/3289275 20:56
21:47 bluescreen joined 23:05 benabik joined 23:55 benabik 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!