Front page | perl.perl5.porters |
Postings from November 2011
On Wed, Nov 23, 2011 at 09:07:36AM -0700, Tom Christiansen wrote:
> Karl Williamson <public@khwilliamson.com> wrote
> on Wed, 23 Nov 2011 07:51:59 MST:
>
> >>>> I don't see why using a \N{} sequence should have any impact on what
> >>>> functions are available to use in my code.
>
> >>> Compare:
> >>>
> >>> $ perl -wE 'binmode STDOUT; warn File::Spec->catfile("foo","bar")'
> >>> Can't locate object method "catfile" via package "File::Spec" (perh<SNIP>
> >>>
> >>> with:
> >>>
> >>> $ perl -wE 'STDOUT->binmode; warn File::Spec->catfile("foo","bar")'
> >>> foo/bar at -e line 1.
>
> >> Sure, I think that's confusing too.
>
> > FWIW, I find this confusing as well, yet this is how things work today:
>
> > $ blead -E '
> > print charnames::viacode(0x41), "\n";
> > package foo {
> > use charnames();
> > }'
> >
> > LATIN CAPITAL LETTER A
>
> > Loading charnames in a different scope, even package, enables
> > use of the functions at a distance.
>
> Ayup.
>
> > Does this fact change anyone's opinion?
>
> It may, but it doesn't change *mine*. This is just how things happen
> in Perl, and there's no way to address it without redesigning the whole
> language from the ground up; see Perl6.
>
> And er, I've actually made intentional use of the sort of thing you show in
> the quoted portion. Similarly with IO:: stuff. It's probably a dumb thing
> to do, but I knew that was how Perl happens to work, so I did it. It's a
> bit embarrassing but there it is. You use what's there.
>
> Perl really isn't spec-based to allow one to adjudge a bug from a feature
> based on a formal spec, so there's a lot of this sort of thing going around.
Right. And this is why I'd like to try to avoid adding more things like
this whenever possible - these are the sorts of things that makes the
language a lot harder to change in the future. Can we even make IO::File
stop loading File::Spec at this point? Someone may be relying on this
autoloading behavior, and their code would break. (To be clear, I think
the answer to this question is "yes", but the potential for breaking
existing code isn't something that should be ignored either.)
-doy
Thread Previous
|
Thread Next