Front page | perl.perl5.porters |
Postings from December 2010
On 13 December 2010 17:47, Abigail <abigail@abigail.be> wrote:
> On Mon, Dec 13, 2010 at 05:26:29PM +0100, demerphq wrote:
>> On 13 December 2010 17:09, Eric Brine <ikegami@adaelis.com> wrote:
>> > On Mon, Dec 13, 2010 at 3:07 AM, Johan Vromans <jvromans@squirrel.nl> wrote:
>> >
>> >> [Quoting Eric Brine, on December 13 2010, 02:51, in "Re: [perl #80628] [P"]
>> >> > On Mon, Dec 13, 2010 at 2:09 AM, Johan Vromans <jvromans@squirrel.nl>
>> >> wrote:
>> >> >
>> >> > > More important is whether it is a keyword, or a compile time constant
>> >> > > like __FILE__ and __LINE__.
>> >> > >
>> >> > > -- Johan
>> >> > >
>> >> >
>> >> > As proposed, it's not constant.
>> >>
>> >> Then please do not call it __SUB__.
>> >>
>> >> What would be the difference with (caller(0))[3] ?
>> >>
>> >
>> > caller[3] doesn't work for anon subs.
>> >
>> >> perl -E"sub { say( (caller(0))[3] ) }->()"
>> > main::__ANON__
>> >
>> > It was specifically proposed that __SUB__ would return a reference in at
>> > least that case.
>>
>> It seen's to me that it MUST return a reference in ALL situations.
>
> I agree. Do one thing in one case, and another in another case doesn't seem
> like a win situation to me. At least, not here. You certainly shouldn't
> have to change the body of a sub if you change it from an anon sub to a
> named one.
And the fact that it not returning a reference would almost certainly
lead to bugs.
For instance the same kind of problem we see here:
$ perl -MCarp -le'sub foo { confess "foo" } sub baz { confess "baz" }
*bar=&foo; *foo=&baz; bar()'
foo at -e line 1
main::foo called at -e line 1
If we returned anything but a code reference I reckon one could cook
up something like this to cause a bunch of edge-case insanity.
(Consider how much worse it gets when you add localized subroutines to
the mix...)
cheers,
Yves
--
perl -Mre=debug -e "/just|another|perl|hacker/"
Thread Previous
|
Thread Next