01:52 sortiz joined 02:13 ugexe joined 06:50 geekosaur joined
dalek ast: a1cd3e7 | usev6++ | S (2 files):
Fudge tests for RT #128031 on JVM
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128031
dalek ast: 985a4a0 | usev6++ | S14-roles/mixin.t:
Fudge test for RT #127916 on JVM
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127916
07:12 RabidGravy joined
masak "it's perhaps most accurate to say that V8 supports compliance with the “continually maintained draft future ECMAScript standard”" -- interesting reading. v8project.blogspot.se/2016/04/es6-e...eyond.html 08:37
09:08 vendethiel joined 10:15 pmurias joined 12:00 masak_ joined 12:01 [Coke]_ joined 12:12 JimmyZ joined
psch how is the long[][] handlers in StaticCodeInfo structured anyway..? 12:48
it's documented as "Map of handlers"
...and how do i get the equivalent of this gist.github.com/peschwa/d61db7a81c...94ee45bc65 for r-m? :S 12:56
i mean, something feels weird to my limited understanding there. first off, the current CF doesn't have a handler at all, but the first caller has one for 'last' 12:57
i mean, i might be misunderstanding that, but the nqp::debugnoop call shouldn't introduce a new CallFrame, should it?
and then there's two more handlers, with what i think are invalid categories 13:04
oh, or they're for more than one category... 13:05
aha. tc.curFrame.caller.curHandler is 342, which is the handler that handles next, last and redo in the gist, i.e. handlers[2] 13:07
oh, the caller of the for is the scope/block that applies the label to the for block, isn't it 13:18
so it's not the debugnoop that adds a frame but it's the for loop that does
dalek p: c459573 | (Pawel Murias)++ | src/vm/js/nqp-runtime/sixmodel.js:
[js] Support nqp::atpos on type objects.
p: 1daaef5 | (Pawel Murias)++ | src/vm/js/RegexCompiler.nqp:
[js] Initialize variable before use.
p: a28876c | (Pawel Murias)++ | src/vm/js/nqp-runtime/nfa.js:
[js] Fix a bug with NQPMu $!cstack breaking the bstack.
p: 0234430 | (Pawel Murias)++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/IOOps.java:
[jvm] Remove commented out code.
psch pmurias++
i totally missed removing that
updated the gist, fwiw 13:34
the way i see it, we're simply not installing a handler when we don't find any of next/last/redo inside the block
instead we should probably also just do it if we find a label 13:35
[Tux] This is Rakudo version 2016.04-53-g1acf805 built on MoarVM version 2016.04 13:37
test 21.940
test-t 13.421
csv-parser 23.385
13:39 RabidGravy joined 13:50 pmurias joined
psch hm, just installing handlers if we have a label apparently isn't enough /o\ 14:26
ohh, wait. for gets compiled as p6for, which desugars to a map call 14:31
14:34 skids joined 14:35 brrt joined
dalek kudo/nom: edd8964 | (Zoffix Znet)++ | src/core/IO/Path.pm:
Fix wrong method names in exception messages for file tests
kudo/nom: b939c4e | lizmat++ | src/core/IO/Path.pm:
Merge pull request #761 from zoffixznet/patch-1

Fix wrong method names in exception messages for file tests
kudo/nom: d7698f3 | lizmat++ | src/core/ (2 files):
Fix for RT #128039
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128039
16:34 geekosaur joined 16:55 pmurias joined 18:16 brrt joined
dalek ast: 6d2545c | usev6++ | S16-io/supply.t:
Fudge test for RT #128041 for rakudo-j
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128041
18:21 dalek joined
dalek p: 1c90b06 | (David Warring)++ | examples/rubyish/rubyish.nqp:
[rubyish] disambiguate '=' && '=='

fixes some parse errors in latest nqp-m
p: 67e04c7 | (David Warring)++ | examples/rubyish/t/hashs.t:
[rubyish] fix hash-key ordering in test suite
p: a6739ba | (David Warring)++ | examples/rubyish/README.md:
[rubyish] prefer MoarVM over Parrot in README.md
19:47 sortiz joined
sortiz \o #perl6 19:47
stmuk_, DBIish update don't require --force, the version was changed. 19:53
ugexe maybe needs to do `zef/panda update` to update the package list with the updated version number 21:07
ugexe needs to document zef auto package-list update config option 21:08
sortiz timotimo, Trying to bisect the spectest failure found nothing. I can't reproduce it. So, or some garbage in my workdir or some heisenbug. 23:17
timotimo oh, could be :\ 23:24
we tend to call these tests "flappers"
it was start.t, right?
sortiz yep 23:26
timotimo yeah, multi-threaded stuff ... 23:27
i'm not 100% sure whether the deadlock bug was due to the new code in jnthn's branch or an old one that's always been there
sortiz I'll keep an eye on it. 23:30
timotimo have you read jnthn's last report? 23:33
sortiz Yes. 23:36
If I understand well, that work hasn't landing yet in MoarVM master, right? 23:38
timotimo correct 23:39
having the frames reference-counted instead of gc-managed used to be a major factor in their speed 23:40
so until the optimization lands that makes frames only gc-managed when they "escape" ... everything that involves frames will be crazy slow
sortiz After read jnthn discoveries, I somewhat surprised that Rakudo works ;-) 23:41
timotimo well, those GC issues are really hard to hit. extremely hard to hit reliably 23:42
unless of course you turn on gc torture testing :)
sortiz Yep, it is clear. 23:43
timotimo and there was already a nice torture testing session like that a year or two ago, and the features that were already there at that point got the benefit of the testing already
of course, those features likely were touched in the mean time, too