I see what you mean, but I don't see that as a requirement. The first reference taking could just as easily return undef. In the odd (strange and rare) case you need to take a reference to a function that hasn't been compiled yet, you could always declare it using a stub.

Perl has long allowed the taking of references to functions before they have been compiled or even declared. So support for that is indeed required until after the decision is made to no longer support that and after the deprecation cycle can happen.

- tye        

Of course. I'm not arguing for such a change. There already exists sufficient tools (defined) to check if a function exists. I'm just challenging the claim that autovivification through \& is necessary to call functions loaded dynamically.

Strange and rare? A simple dispatch table illustrates taking a reference before a function is seen, as the table typically comes before subroutines.

my %OPS = (
     frob => \&frobnicate,
     knob => \&knobnicate,
);

sub frobnicate { ... }
sub knobnicate { ... }
[download]

As for a reference evaluating to undef, the concept seems wrong. If a reference can't be taken it should be an error; if it can, you should get a reference. Are there any examples of this behavior in Perl now?

No, it doesn't. The function gets created by sub before \& is executed.

print "a:", exists(&frobnicate)?1:0, "\n";             # 1

my %OPS = (
     frob => \&frobnicate,
     knob => \&knobnicate,
);

BEGIN { print "b:", exists(&frobnicate)?1:0, "\n"; }   # 0

sub frobnicate { '...' }
sub knobnicate { '...' }
[download]