Front page | perl.perl5.porters |
Postings from March 2012
On Wed, Mar 28, 2012 at 04:42:02PM -0400, Matthew Horsfall (alh) wrote:
> On Wed, Feb 29, 2012 at 12:40 PM, David Cantrell <david@cantrell.org.uk> wrote:
> > On Thu, Feb 16, 2012 at 05:06:51PM +0000, I wrote:
> >> +It is up to this method to store the list of attributes if they will be
> >> +needed later, as well as checking for any errors. In this example there
> >> +are no error conditions, so we just store:
> >> +
> >> + my %attrs;
> >> + sub MODIFY_CODE_ATTRIBUTES {
> >> + my($package, $subref, @attrs) = @_;
> >> + $attrs{ refaddr $subref } = \@attrs;
> >> + return;
> >> + }
> > Turns out that blindly looking at refaddr($subref) doesn't always Do The
> > Right Thing. This code gets called when the subroutine with attributes
> > is being compiled, and if it's a closure you sometimes (maybe always?)
> > get a different refaddr at this point than the subroutine ends up with
> > after it's been compiled. I have no idea off the top of my head how to
> > fix this to take that into account.
> >
> > Also I seem to remember there being some Argument recently about whether
> > refaddr() should be in core or not but forget what the final decision
> > was. In this case (modulo the closure stuff above) simply stringifying
> > the subref does the job, so refaddr can be cut out.
>
> Should the docs be updated again then? The original patch is still
> applied and this last email seems to indicate it needs to be changed.
Updating to get rid of refaddr, yes. As for how to deal with closures I
don't know what would be best. The doco definitely needs this little
bit of explication for the simple non-closure case.
--
David Cantrell | London Perl Mongers Deputy Chief Heretic
When a man is tired of London, he is tired of life
-- Samuel Johnson
Thread Previous
|
Thread Next