github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
samcv .tell dogbert11 i'm not getting those valgrind errors you are 06:39
yoleaux samcv: I'll pass your message to dogbert11.
samcv i will try compiling with gcc instead of clang and then try again 06:41
there's an error in the code though 06:44
it should just be optimized out i'd think. but there's some extra variable definitions that shouldn't be there 06:53
Geth MoarVM: ec8a240e32 | (Samantha McVey)++ | src/strings/utf16.c
Remove some unused variables in utf16 decode

Should hopefully resolve issue #1030 which gave valgrind errors due to uninitialized variables.
07:10
brrt \o 14:20
nwc10 o/
jnthn o/ 14:26
[Coke] o) 14:32
timotimo ooh, the variables were entirely unused, that's why in more optimized builds there was no null pointer deref 14:38
dang, i could have thought of that :D
dogbert2_ seems a bit dangerous to assume that the dangerous/broken code will be optimized away though 14:40
timotimo if null pointer deref is deemed undefined behaviour, then it's absolutely fine for the compiler to do whatever it feels like 14:42
[Coke] is there a way to get a 64bit build of moarvm on windows? 15:14
timotimo aren't the only builds on windows we have 64bit 15:15
?
[Coke] just noticed my local build is reporting 32 bit in taskmgr. 15:16
timotimo oh
timotimo not actually sure how to work msvc, but i'd start with the appveyor file 15:20
brrt I'd also expect the windows build to be 64 bit... otherwise the JIT won't work :-) 15:28
timotimo i'm glad to see that coverage is still being uploaded, but of course it's only using the nqp test suite, so no info about rakudo's extops 15:51
timotimo hum 16:21
the coverage files i'm getting are 0 bytes big 16:22
AlexDaniel nine: ooooohh, hmmm 17:19
I wonder if resolving this will help RT#131003
synopsebot RT#131003 [open]: rt.perl.org/Ticket/Display.html?id=131003 [SEVERE][SEGV] Heap corruption when using Gumbo 17:19
AlexDaniel ah no, that one can be reproduced with MVM_SPESH_DISABLE=1 17:24
ok then, maybe next year
:)
samcv dogbert2_, well. it's more a thing of the variable being defined in an outer context. but not used in outer scope. but it's defined again in an inner scope with same name 18:23
so i don't think it should be used... unless i'm missing something. which is possible
dogbert11 samcv: yes the valgrind errors only showed up if MoarVM was built with --no-optimize 20:13
yoleaux 06:39Z <samcv> dogbert11: i'm not getting those valgrind errors you are
dogbert11 anyway the problem seems to have been resolved, samcv++ 20:14
samcv ok great :) 20:50
night all o/