github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
Geth MoarVM: ugexe++ created pull request #1082:
Add fileno_fh legojit and exprjit template
00:05
timotimo ah i found the cause of the 0% jittage 00:08
Geth MoarVM: 38252329ed | (Timo Paulssen)++ | src/profiler/heapsnapshot.c
Heapsnapshot: properly handle empty debug name

the code was expecting empty debug names to be returned as null pointers, but the debug name API returns those as empty strings instead so that no null pointer checks are necessary.
00:15
MoarVM: bacfcf96e2 | (Timo Paulssen)++ | src/6model/reprs/CStruct.c
Heapsnapshot: Calculate CStruct unmanaged size
MoarVM: 3934160407 | (Timo Paulssen)++ | src/jit/core_templates.expr
exprjit/profiler: fix enters not showing up as jitted
timotimo MasterDuke: ^- that snuck through, i should probably have noticed that sooner :) 00:16
ah, dang, it had just made it into the release 00:17
Geth MoarVM: ugexe++ created pull request #1083:
Add decodeconf legojit
01:40
MoarVM: ugexe++ created pull request #1084:
Add istty_fh legojit
02:05
ugexe fwiw i can create one PR if requested. i didnt originally plan on going through this many 02:23
i jitted backtrace and backtrace strings, but then using `from-json` (which usually generates 1 deprecation) generates 2 deprecations if i call from-json 10000 times instead of 1000 03:20
Block <unit> (from unknown) seen at: 03:21
Sub from-json (from GLOBAL) seen at:
both with the same message: "Please use JSON::Fast, JSON::Tiny or JSON::Pretty from modules.perl6.org/ instead." 03:22
ah i take it back, that happens even without that change 03:28
still, wonder what causes that 03:29
MasterDuke timotimo++ MasterDuke-- 03:45
Geth MoarVM: ugexe++ created pull request #1085:
Add backtrace and backtracestrings lego and exprjit template
03:50
dogbert17 it's possible to make t/04-nativecall/06-struct.t crash ~100% of the time if the nursery is made smaller 15:36
perhaps something for nine? gist.github.com/dogbert17/39020911...5659ccc38e 15:37
timotimo that's the "CArray embedded in CStruct will cause an incorrect free" thing 15:41
you can tell by looking at what frees and what allocated; CArray tries to free, CStruct did the allocation of the pointer that's 4 bytes past the beginning 15:42
dogbert17 a known problem? 15:43
timotimo aye 15:44
i should really have started on this much earlier
well, that makes it sound like i've actually already started on it 15:45
dogbert17 :)
is it an easy or difficult fix? 15:46
timotimo well, a simple fix shouldn't be all that hard, but it'd be great to have something Very Good instead 15:48
dogbert17 and that is Very Hard I presume 15:49
timotimo implementing it would probably still be easy. just not coming up with what the right way is :) 15:52
dogbert17 sounds a bit like finding a good name to a variable or method, sometimes that can be incredibly difficult 16:02
timotimo aye 16:08
i like how the :$foo shortcut sometimes makes it trivial to come up with "the right name" for a thing in your code 16:09
because you can more easily pass it that way