> Will the proposed new want operator make writing wrapper
> subroutines which recreate their own context for the
> subroutine they are wrapping harder?
I don't see why.
> For example how would the following perl 5 subroutine
> be rewritten under perl 6 with the want operator?
>
> sub wrapper {
> my (@ret,$ret);
>
> lock_resource();
>
> if (wantarray) {
> @ret = &wrapped;
> } elsif (defined wantarray) {
> $ret = &wrapped;
> } else {
> &wrapped;
> }
>
> unlock_resource();
>
> if (wantarray) {
> return @ret;
> } elsif (defined wantarray) {
> return $ret;
> } else {
> return;
> }
sub wrapper {
my (@ret,$ret);
lock_resource();
switch ([want]) {
case 'LIST' { @ret = &wrapped }
case 'SCALAR' { $ret = &wrapped }
else { &wrapped }
}
unlock_resource();
switch ([want]) {
case 'LIST' { return @ret }
case 'SCALAR' { return $ret }
else { return }
}
}
Alternnatively, the return section could be just:
return want 'LIST' ? @ret
: want 'SCALAR' ? $ret
: undef;
> Presumably wrapped() currently does one of 3 different
> things according to the value of wantarray but might be
> enhanced to do one of up to 21 different things according
> to the value of want.
True, but most of those other ways are either irrelevant to this example
(i.e. make no difference to the existing behaviours) or they are sub-cases
of the existing cases (which one might or might nor choose to optimize).
Damian
