He probably means something hypothetical...

I was actually thinking about this seriously, and I would likely do something as:

sub get_sub_by_name {
    my ($pkg, $name) = @_ ;
    no strict 'refs' ;
    if ( exists ${$pkg . '::'}{$name} ) {
        if ( ref ${$pkg . '::'}{$name} eq 'CODE' ) {
            return *{$pkg . '::' . $name} ;
        }
    } ;
}
[download]

Maybe it can be written more pretty, and I don't know if any odd-cases are trapped (e.g. what about __ANON__)

edit: Comes to think of it, maybe the line 'exists' is not even needed?

I'm afraid your hypothetical code for his hypothetical function is ignoring how typeglobs work.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery FootballPerl is like chess, only without the dice}

Oh yeah, haha, what have I been smoking!

I'll probably stick with this then:

sub get_sub_by_name {
    my ($pkg, $name) = @_ ;
    no strict 'refs' ;
    if ( defined &{$pkg . '::'. $name} ) {
        return \&{$pkg . '::'. $name} ;
    }
}
[download]

Some code that gets a reference to the sub by name. What that code is doesn't matter, though it could be as simple as \&{ $pkg . '::' . $name }.

> could be as simple as \&{ $pkg . :: . $name }.

nope, this will autovivify a missing function glob $name and return a broken coderef.

try again.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery FootballPerl is like chess, only without the dice}

You're missing the point. I'm not proposing a solution; I'm saying that yours is ludicrous. It makes absolutely no sense to use

my $sub =
   @{ get_array_by_name($pkg, 'ISA') }
      ? get_sub_by_name($pkg, $name)
      : $pkg->can($name);
[download]

instead of

my $sub = get_sub_by_name($pkg, $name);
[download]

The implementation of get_sub_by_name isn't relevant to that point.