nwc10 good *, #moarvm 08:20
jnthn o/ nwc10 08:21
pmurias jnthn: when do we increase the serialization format version? 12:21
jnthn pmurias: When making any change that would render a version of MoarVM that doesn't know about a change unable to successfully/completely deserialize things 12:24
pmurias jnthn: thank you, had a disagreement about my "Serialize hll role" pull request 12:33
jnthn pmurias: Yeah, keep meaning to review that :) 12:35
pmurias I just found an empty else in it :(
(fixed that) 12:38
brrt ohai #moarvm 13:49
hoelzro a MoarVM frame may or may not be a callable, right? it could just be something like an if $cond { ... } block? 16:33
I suppose my question is - is there a way to tell if a frame is the top level frame of a method vs an arbitrary lexical frame? 16:44
jnthn hoelzro: Not in MoarVM 19:12
But Moar shouldn't really care
At QAST level, the blocktype is immediate for things like if $cond { ... }
hoelzro ah ha 19:14
jnthn: I'm just digging into that "nextsame looking for $*DISPATCHER in its caller" behavior I saw
wondering if p6finddispatcher should stop crawling up the frame stack once it hits the sub boundary
jnthn Hm. Or maybe we just need to install a $*DISPATCHER in all routines that "blocks" the other one 19:16
hoelzro ah, that could work too 19:17
at first blush, that doesn't sound terribly performant, but I'm guessing it's ok if you suggested it =)
jnthn It's just about free
Since all lexicals are lazily vivified :)
hoelzro has much to learn 19:18
jnthn I forget exactly the details around this stuff :)
(Haven't had to touch it for a long while...)

Powered by Cro and the Raku® Programming Language.

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

Thank you!