Like I was saying: > If the two [multis] are defined in different modules I suspect you > could get some strange behavior that would be hard to debug. > It's not always immediately obvious in what order two things were > defined.
Here's an example, scattered across four files, which I've also pushed out to github: https://github.com/doomvox/raku-study/blob/main/md_exp/ The deal here is one script treats "godzilla" as a Hero, the other as a Monster, and the only difference is the order of two "use" lines: --- md_exp/lib/Speak/Monster.rakumod: module Speak::Monster { my @monsters = < godzilla gammera ghidora golem >; subset Monster of Str where { $_ eq any( @monsters ) }; multi sub speak (Monster $name) is export { say "The monster, $name roars!"; } } --- md_exp/lib/Speak/Hero.rakumod: module Speak::Hero { my @heroes = < godzilla beowulf ultraman inframan >; subset Hero of Str where { $_ eq any( @heroes ) }; multi sub speak (Hero $name) is export { say "The hero, $name shouts!"; } } --- md_exp/bin/multidispatch_overlapping_subsets_use_order_dependency-a.raku use lib $*PROGRAM.parent(2).add("lib"); { say "a: Hero Monster trial run"; use Speak::Hero; use Speak::Monster; speak('godzilla'); ## The hero, godzilla shouts! } --- md_exp/bin/multidispatch_overlapping_subsets_use_order_dependency-b.raku use lib $*PROGRAM.parent(2).add("lib"); { say "a: Monster Hero trial run"; use Speak::Monster; use Speak::Hero; speak('godzilla'); ## The monster, godzilla roars! } --- But the situation is even worse than this: the order-of-definition doesn't depend on the order of the use lines in the block of code in front of you: in a large code base you might have use lines scattered in different places, and the order-of-definition is the order in which raku first sees the use lines.
