japhb Out of curiosity, when the op in question just resolves to a standard C function (like the FP math functions), why not just jit them all? Why add one at a time? 00:39
timotimo driven by need in my case 00:41
feel free to add more 00:42
imo having test code helps
japhb Well, sure.
timotimo the big integer ops need a little bit of transforming before they can be put into the jit as is 00:43
hoelzro ahoy #moarvm 18:09
so I tried adding some printf debugging statement to moduleloader 18:11
nothing appears to be different =/
(btw, this is regarding rt.perl.org/Public/Bug/Display.html?id=122773)
my only assumption is that the internal dictionary for the Example stash is being cleared or set to a new dictionary with the lazy deserialization stuff 18:12
lizmat hoelzro: where did you do that? 18:32
hoelzro lizmat: are you asking if I pushed that code somewhere? 18:43
or where in the source I put the debugging stuff?
lizmat no, just wondering where in the source you put that
indeed
hoelzro src/Perl6/ModuleLoader.nqp
and src/6model/serialization.c
lizmat ok, so maybe this question more belonged in #perl6 :-) 18:49
fwiw, after the next release, I'll be diving into the module loader code and finally get S11 and S22 up to spec 18:50
timotimo the lazy deserialisation is at a different level than that 19:06
hoelzro I agree 21:08

Powered by Cro and the Raku® Programming Language.

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

Thank you!