One problem with using dynamic scoping is that you've opened up the possibility that the same variable could be used somewhere else for a different purpose. That is the advantage to using lexicals.
Another solution is to use ReleaseAction to do cleanup when you exit the sub.
sub outer {
my $helper;
$helper = sub {
...
};
my $delay_cleanup = on_release {$helper = undef};
$helper->(@_);
}
Effectively it does the same thing as the weaken solution, but the code is a little more explicit about what is happening, when. I'd personally lean towards this solution.
And a final option. Larry thinks that using dynamic scoping on lexical variables is too confusing to allow. But he does allow it for data structures. Which leads to the convoluted:
sub outer {
my @helper;
local $helper[0] = sub {
...
};
$helper[0]->(@_);
}
I wouldn't suggest this solution for fear of the psychopathic maintenance programmer demanding to know why this works.
But every year or two I wind up using this fact in a recursive function to detect and track down potential deep recursion bugs. For instance I'm recursing through a set of nodes and I'll do something like this:
{
my %in_node;
sub something_recursive {
my $node = shift;
if ($in_node{$node->{name}}) {
confess("Can't access $node->{name} while accessing $node->{name
+}");
}
local $in_node{$node->{name}} = 1;
...
}
}