Re: Anonymous multi-subs

On Wed, Jun 24, 2015 at 7:17 PM, Brent Laabs <[email protected]> wrote:
I'll just note that you can fake anon multi subs with lexical subs like
this:

my $sub = do {
    proto foo (|) { * }
    multi foo (Int $x) { $x + 1 }
    multi foo (Str $y) { $y ~ 'a' }

    &foo;
}

say $sub("hello");

I like that, and I suspect it could be made generic (a routine that takes a
list of subs, returning a multi-sub) without too much difficulty.

The sub there is still named "foo" as attested by $sub.name, but isn't
> available under that name outside of the do block.
>

I've found named lambdas very useful when debugging (in Javascript). Having
$sub.name available is a feature as far as I'm concerned!


On Wed, Jun 24, 2015 at 5:27 PM, Jon Lang <[email protected]> wrote:

 Or, in the body, be able to examine the actual signature passed in and
> decide what to do based on that,  That [i]could[/i] be done using a
> given/when structure, which would be equivalent to a list of signature/body
> pairs.
>
>
That works, but it means re-doing what the dispatcher already does. Nicer
to hook into existing machinery.

Shortly after writing my last message I realized that a "*signature*, *body
*pair" is exactly a sub, and that what surprises me is that there isn't
already a constructor that returns a multi-sub, given a list of subs.

S06 says "proto is a generic wrapper around the dispatch to the multis" and
"A proto is a generic dispatcher, which any given scope with a unique
candidate list will instantiate into a dispatch routine" ... that's the
machinery to hook into. I'm thinking of them as roles or types, when really
they're attributes of Routine.