PPC0034: refalias in subroutine signatures

[dakkar]

as discussed on p5p: allow sub ($one, \$two, \@three, \%four ||={}) and similar constructrs

[leonerd]

I think overall this seems fine.

One thing that isn't made super obvious from the doc is the interaction between mutations of referred variables, and defaulting expressions. I suppose you'd get something like a new temporary container that is mutable, but gets thrown away once the call is finished. That's likely how it would fall out of the implementation.

I'd be happy to accept this, but will wait for any comment from @Leont or @ap as well first.

[dakkar]

I feel that any clarification on the behaviour of defaulting expressions is outside this PPC, since it's not specific to ref-aliasing.

I agree that the existing documentation does not really explain how:

sub foo($a=[]) { push $a->@*, 1, 2, 3; return $a }

works (it works the way you guessed: each call gets a fresh arrayref)

[leonerd]

We discussed this in the PSC meeting today, and it seems good to merge. We can continue iterating further from there.