Dan Sugalski wrote:
[snip]
> All of these--method lookup, property lookup, attribute lookup--may
> be intercepted, and all may have a fallback method that's called if
> the 'proper' lookup fails.
>
> I think this about covers it. If there's missing semantics, and I
> expect I missed something, let's get it out and add it. Once we've
> got it pretty much nailed I'll spec out the interface and we can
> start in on the implementation.
For the sake of completeness, I would like to ask, what are the
semantics of multiple inheritance?
Do we *only* natively offer a way of doing perl6 method dispatch
(whatever that may be), and any other type of method dispatch else has
to be emulated using a subroutine written in bytecode?
Or, do we offer multiple ways to look methods up... one way for each
language that we're compiling? So that we could compile C++ to parrot,
or compile perl(5|6) to parrot, or compile python to parrot, etc; using
a different operator (or one operator, with a parameter indicating the
language used for method dispatch) for method calls for each language?
If so, do we offer a way of extending parrot to allow other method
lookup schemes?
It would be sad if we only offered perl5 and perl6 method dispatch using
a single op each, and had to use multiple ops each to emulate other
languages' method dispatch techniques.
--
$;=qq qJ,krleahciPhueerarsintoitq;sub __{0 &&
my$__;s ee substr$;,$,&&++$__%$,--,1,qq;;;ee;
$__>2&&&__}$,=22+$;=~y yiy y;__ while$;;print
