Yes, that is one difference, but I find code that makes use of that difference to be a bit of a problem. I'd probably write HI_POST_PI()+5 if I noticed.
The difference that I'm talking about is that subroutines with a () prototype that also do nothing but return a value become compile-time constants. Perl can replace instances of calls to these with the constant returned at compile time, allowing lots of interesting tricks:
> Perl -MO=Deparse -w
use strict;
sub DEBUG() { 0 }
sub SymRef() { "My::Package::hash" }
sub PI() { 3.141592 }
if( DEBUG ) {
warn "Okay, we are debugging...\n";
do_lots_of_debugging_stuff();
}
$My::Package::hash{key}= "value";
print SymRef->{key},$/;
print PI()+5,$/;
__END__
sub DEBUG () {
0;
}
sub SymRef () {
'My::Package::hash';
}
sub PI () {
3.141592;
}
'???';
$My::Package::hash{'key'} = 'value';
print $My::Package::hash{'key'}, $/;
print 8.141592, $/;
- syntax OK
Without the () prototypes we'd get:
sub DEBUG {
0;
}
sub SymRef {
'My::Package::hash';
}
sub PI {
3.141592;
}
if (DEBUG ) {
warn "Okay, we are debugging...\n";
do_lots_of_debugging_stuff ;
}
$My::Package::hash{'key'} = 'value';
print SymRef()->{'key'}, $/;
print PI() + 5, $/;
- syntax OK
and, if we ran the code, we'd get:
Can't use string ("My::Package::hash") as a HASH ref
while "strict refs" in use
-
tye
(but my friends call me "Tye")