I made a comment on your blog which is awaiting moderation. Anyways here it is for people on the mailing list:
I agree that being able to compose dispatchers and candidates procedurally is very useful. And you can sorta do it. Check out: https://github.com/LLFourn/p6-CompUnit-Util#dispatcher-manipulation especially this line: https://github.com/LLFourn/p6-CompUnit-Util/blob/master/lib/CompUnit/Util.pm6#L233 where I have to create a proto. This works but there is a serialization bug which breaks precompilation if it's done at compile time. but doing it at runtime should work: sub create-proto($name) { my $proto = (my proto anon (|) {*}).clone; $proto.set_name($name); return $proto; } my $proto = create-proto('foo'); $proto.add_dispatchee(anon sub candidate1('one') { 'one' }); $proto.add_dispatchee(anon sub candidate2('two') { 'two' }); say $proto.('one'); say $proto.perl; .say for $proto.candidates; The above will work but it's quite fragile. Try assigning the sub to a variable first and then .add_dispatchee (it will die). It definitely needs a nicer interface. The design docs say that protos should have .push and mentions that you should be able to do Proto.new. https://design.perl6.org/S06.html#Introspection On Mon, Feb 15, 2016 at 6:36 AM yary <not....@gmail.com> wrote: > Back in June of last year some discussion about multi-subs got me > thinking and posting about anonymous proto/multi routines here. It's > been bubbling in the back of my mind since then, and as my Valentine > to the language, I've posted my thoughts at > > http://blogs.perl.org/users/yary/2016/02/apropos-proto-perl6c-multi-thoughts.html > > It's a more coherent (I hope!) follow-up to > https://www.mail-archive.com/perl6-language@perl.org/msg34937.html > > Would appreciate feedback on the merits or perils of exposing a way > for creating a truly anonymous multi. It's something the design spec > isn't explicit on. > > -y >
