On Fri, Jul 11, 2008 at 03:27:26PM +0200, TSa wrote:
> HaloO,
>
> Patrick R. Michaud wrote:
>> S29 doesn't show a 'sort' method defined on block/closure
>> invocants... should there be?
>
> I doubt that. And to my eyes it looks funny. Only real block
> methods should be useful and since the class is mostly known
> at parse time unapplicable methods should be a compile error.
>
> my &f = { $^a <=> $^b }.assuming($^a = 3);
>
> say f(3); # prints 0
>
> Would that be valid? I mean the usuage of automatic variables
> in the assuming method?
No, that could not work, because $^a = 3 would be in an rvalue context
and refer to the block around the entire statement as a different
parameter. Setting a named parameter with .assuming must use named
argument notation:
my &f = { $^a <=> $^b }.assuming(:a(3));
Also, it may well be that f(3) will say Order::Same rather than 0. :)
Larry
