Front page | perl.perl5.porters |
Postings from June 2011
On Mon Dec 13 13:35:19 2010, jesse wrote:
>
>
>
> On Mon, Dec 13, 2010 at 03:17:11PM -0500, Jesse Vincent wrote:
> > I'd rather this functionality live on CPAN until there's a clear
> consensus that it
> > should be in core and about what it should be named.
>
> I somewhat misphrased this - We're very close to the
> contentious-feature-freeze for 5.14. And this is clearly contentious.
> I'd much rather it get better sorted out before we take it into blead.
> To me, it makes sense that it get beaten around a bit before we think
> about the eventual result for 5.16.
OK, now that we’re past that, let me address some objections that came up:
1) That __SUB__ is not a compile-time constant like its predecessors.
That __PACKAGE__ and __LINE__ are compile-time constants is an
implementation detail.
__PACKAGE__’s being a compile-time directive actually causes a bug in
some edge cases, so that is not set in stone.
__SUB__ *can* be a compile-time constant in subroutines that do not
close over variables. That optimisation just hasn’t been written yet.
2) That this should be a CPAN module, not part of core.
One reason I think we should have this in core is that JavaScript
supports it (arguments.callee), which makes Perl look bad.
And the symmetry with package/__PACKAGE__ is nice:
package foo { ... my $pack = __PACKAGE__ ... }
sub foo { ... my $sub = __SUB__ ... }
Thread Next