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.
Nicholas jnthn1: surely you are jnthn6 as you are connected on IPv6? :-) 09:10
jnthn1 Sigh, it appears somebody already registered jnthn here :/ 09:11
MasterDuke55 jnthn1: i left a message for jnthn a couple days ago on freenode, but i'm guessing you didn't get it colabti.org/irclogger/irclogger_lo...05-22#l301 09:16
jnthn1 MasterDuke: Will review it, though gotta go and do some probably time-consuming administrative errand now...bbl 09:20
MasterDuke thanks, no particular hurry for that one 09:21
@lizmat this just on HEAD on my local copy of the logs server: 14:57
raku(31555,0x700002c7d000) malloc: *** error for object 0x7f98d87cd7c0: pointer being realloc'd was not allocated
raku(31555,0x700002c7d000) malloc: *** set a breakpoint in malloc_error_break to debug
just for reference 14:58
Nicholas So we have to keep talking, to try to recreate the test case? That's what you're saying? :-) 15:33
@lizmat he, no, just noticing there's a gremlin in there still 15:34
haven't been able to repro
nine You probably don't have a core dump? 15:35
@lizmat Nicholas: would you argue that's a typical across thread hash access issue >?
if so, then I'm probably to blame
Nicholas it's possible, for a large hash
@lizmat nine: no core dump, no
Nicholas for a smaller hash, you'd confuse the FSA, which might mess things up in a different way 15:36
but it will all end up with undefined behaviour
(and much wailing and gnashing of teeth)
@lizmat hehe 15:37
should I mention these events here or only if I can repro ?
nine lizmat: please mention them. Maybe someone is working on something related or has seen something related and picks up a further clue that way. We've got channel volume well under control :) 16:07
jnthn1 Curious find. /<&foo()>/ compiles into CALL_SUBRULE. /<&foo>/ compiles into INTERPOLATE_ASSERTION. I think this was an accident. But the latter is likely a good bit less efficient. 16:51
Fixed in RakuAST (TM)
MasterDuke i don't know CALL_SUBRULE, but generally i say the fewer calls to INTERPOLATE_ASSERTION the better 18:45
moon-child maybe worthwhile to run under rr in the future, so easier to catch it if it happens again? 22:57

Powered by Cro and the Raku® Programming Language.

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

Thank you!