Piers posed the following puzzles:
> @ary[0] = foo() # scalar
Yes.
> @ary[1,2] = foo() # list context
Yes.
> @bar = 1;
> @ary[@bar] = foo() # ? probably list or maybe scalar...
List. With an explicit array as index, it's definitely a (one-element) slice.
> @bar = (1,2);
> @ary[@bar] = foo() # list?
List. Same reason as previous.
> @bar is constant = 1;
> @ary[@bar] = foo() # We know at compile time there's only one thing in
> # @bar. Does that mean foo() is in a scalar context
> # now?
List. Same reason as previous.
> sub a_scalar { 1 };
> sub an_array { my @a = (1,2) }
> sub context { wantarray ? (1,2) : 1 }
sub context { want 'LIST' ?? (1,2) :: 1 }
> @ary[a_scalar()] = foo() # ???
> @ary[an_array()] = foo() # ???
> @ary[context()] = foo() # ???
I'm not sure Larry has decided how this will be handled yet.
If forced to guess, I'd guess the square brackets will impose
scalar context on &a_scalar, &an_array, and &context, making
the indexing operation a scalar look-up (not a slice), so each
call to &foo is in scalar context.
> At around this point, Dan was heard to say 'Mommy, make the bad man go
> away!'
Please don't scare the new Perl serf, Piers.
> And, just for laughs:
>
> $ref = [1,2];
> @ary[$ref] = foo(); # probably a syntax error
Unlikely. Scalar context, I'd expect
(though Larry would have to confirm that).
Damian
