Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
JimmyZ nine: github.com/MoarVM/MoarVM/pull/1712 is it useful? 03:52
looks like it is replaced by github.com/Rakudo/Rakudo/commit/87...74d5012933 03:54
JimmyZ by github.com/Rakudo/Rakudo/commit/63...b4a0690795 04:24
nine JimmyZ: it's currently not used. However I can only say it definitively once I'm through with RakuAST and know that the alternative really covers all places. 06:35
tellable6 nine, I'll pass your message to JimmyZ
timo OK, I think the problem with the misaligned access is that we're inappropriately casting to MVMP6intBody. when we create a P6Opaque, we're not doing the equivalent of what a C compiler would do when embedding an actual MVMP6intBody, we "just" have the int of the size that we want in there, and the MVMP6intBody has a union of all the different int types, so the alignment constraints for the struct 07:56
are the greatest of the alternatives
nine Hey party people! I know making stuff faster is super fun and all. Could we maybe however shift the focus on stability for a bit? We're rapidly approaching the bad old times when we had to run spectests several times to see wheter those failures are due to our own change or just random crashes :/ 11:16
s/on stability/to stability/
timo you call me "party people" yet you have never offered me a silly party hat ;_; 11:31
MasterDuke's ubsan investigation has been for improving reliability I'd claim 11:34
if you like, you can point me at an issue or two, or some CI runs with what looks like random crashes, maybe I can figure some of that out? 11:35
... if i can somehow repro it locally, that is :| 11:38
jdv flappers are up? 14:05
does there seem to be any particular ways to excite them more? 14:06
lizmat Needing to precomp stuff, is one way 14:07
timo so can we try to torture-test that by running only the spec tests that use something that needs to be precompiled and run it over and over while deleting the precomp after every full run? or every other full run maybe 14:57
lizmat yes, that would be one approach I'd sya 16:16
*say
Nicholas I have a sun hat today. I'm not sure if that counts as a party hat 16:17
and I'm about to back out into the sun. Today is strange. -4 overnight, +20 now 16:18
lizmat yeah.. I think I just go my first sun burn of the year :-) 16:25
[Coke] it's warmed up about 30deg F here in the past few days. 16:28
been hovering at -8C for weeks
jdv its 47f/9c in the nyc. nice out. 16:30
well, 8c
Nicholas: ghr was you, right? 16:31
Nicholas ght? 17:03
jdv great hash refactor 17:42
Nicholas oh yes. "I deny everything" 18:36
lizmat "I have no active memories of that" :-) 18:45
jdv ha:) 19:01
very good! 19:02
lizmat it's one of the most well known lines from our former PM Mark Rutte, now head of NATO 19:03

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!