nine This is...odd. I've just managed for the first time to get a version up and running that actually eliminates the goto at the end of an inlined function instead of just replacing it by a no_op. The key was to move all 3 annotations (INLINE_END, FH_END and FH_GOTO) to the next instruction instead of the previous on removing the goto. 13:06
nine But moving INLINE_END and FH_END onto a non-inlined instruction obviously makes no sense 14:29
jnthn Are you sure it's not just FH_END that is the issue? 14:35
nine I can try 14:38
jnthn: actually FH_END doesn't seem to matter. INLINE_END has to go on the next instruction instead of the previous 14:42
nine With that rakudo's build succeeds, make test is successful and only a couple of spectests fail with dubious "Two terms in a row" errors. Would be nicer if it actually made sense though. 14:50
jnthn Hm, I wonder if FH_END is a post or pre annotation 15:27
nine What does that mean? 16:38
timotimo whether the instruction it's attached to is considered inside or outside the area
samcv good * 19:02
lizmat samcv o/ 19:42

Powered by Cro and the Raku® Programming Language.

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

Thank you!