Have you ever wanted to redefine a subroutine while your program is running? It's possible -- and not involving closures or other manipulations. Here's a little demonstration.

Note a few things:

• First, we define a subroutine called my_sub
• Next, we define a scalar containing the new subroutine -- this can come from anywhere. It's only hardcoded here for demonstration purposes.
• We also define a scalar called my_sub just to show that we're not stomping all over *everything* in the symbol table.
• The real magic is in the eval statement -- we assign the interpolated (and then compiled!) subroutine in $soft_sub anonymously to $sub_ref!
• After a couple of demonstrations of &my_sub and $my_sub, we undefine &my_sub using symbol table access.
• Next, we use typeglob access to assign our anonymous new subroutine to the CODE container of *my_sub.
• Finally, we call the new subroutine and then demonstrate that $my_sub is still what it was.

Note that this can be a security risk, but it may come in handy someday.

#!/usr/bin/perl -w

use strict;

sub my_sub {
    print "This is the first subroutine called my_sub\n";
}

my $soft_sub = 'sub {
    print "This is the redefined subroutine.\\n";
    };';

my $my_sub = "The scalar with the same name.\n";
my $sub_ref;

eval "\$sub_ref = $soft_sub";

my_sub();
print "$my_sub\n";

undef &{ *my_sub{CODE} };
*my_sub = $sub_ref;

my_sub();
print $my_sub;
[download]