brrt had so much to do yesterday i forgot the solution i had cooked up :-( 07:55
nwc10 this "work" thing is overrated 08:24
brrt :-( 08:26
yes....
psch jnthn: can you look at moar PR #416? github.com/MoarVM/MoarVM/pull/416 10:54
jnthn Hm, that error could be a bit clearer/explanatory perhaps 11:06
It'd be nice if it included the type name too
psch right, as commented, the nqp-j equivalent has the class name, but i'm not sure how to do that on moar
like, we don't have the convenient Ops.typeName afaict 11:07
about explanatory... well, the test i pushed only checks the message for "no fields"
i suppose we could explain why no fields are a bad thing for a CStruct 11:08
jnthn Commented on the ticket
Well, commit 11:09
psch right, so that's to be a sprintf with %s for st->debug_name or so i gather? 11:10
jnthn MVM_throw_exception_adhoc works printf-y :)
So just pass it as an extra arg
psch ah, okay
jnthn It's C but we try to make it at least a tiny bit convenient :) 11:11
lunch, bbiab 11:14
nebuchadnezzar rakudo NativeCall tests fail on several architectures lists.alioth.debian.org/pipermail/...00970.html 22:40
when I try with a MoarVM build against libffi, I have some failure for amd64 and i386: lists.alioth.debian.org/pipermail/...00971.html 22:41
jnthn i386 surprises me a bit 22:42
libffi fails don't in so far as it's had a lot less testing
I'm suspecting either alignment or endianness

Powered by Cro and the Raku® Programming Language.

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

Thank you!