Your CLEAR function does too much work, and has a misleading statement.

sub CLEAR {
    my $self = shift;
    foreach ( @$self ) {
        foreach my $key ( keys %$_ ) {
            delete $_->{ $key }
        }
    }
    $self = [ { } ];
}
[download]

I would just write:

sub CLEAR {
  my $self = shift;
  @$self = ( {} );
}
[download]

or the more terse

sub CLEAR {
  $_[0] = [ {} ];
}
[download]

The problem with assigning to just $self is that it doesn't modify the data referred to by $self. And you're taking a long and arduous approach to clearing those hashes.

My other remark is about your push_hash() and pop_hash() functions. Why not prototype them as (\%), and then export them? This will allow you to do:

use Carp;

sub push_hash (\%) {
  my $href = shift;
  my $obj = tied %$href;

  croak "must be a Tie::Hash::Stack tied hash"
    unless $obj and $obj->isa('Tie::Hash::Stack');

  push @$obj, {};
}


sub pop_hash (\%) {
  my $href = shift;
  my $obj = tied %$href;

  croak "must be a Tie::Hash::Stack tied hash"
    unless $obj and $obj->isa('Tie::Hash::Stack');

  pop @$obj;
}


# and in your program...

tie my(%stash), 'Tie::Hash::Stack';

# ...
push_hash %stash;
# ...
[download]

Update: Hmm... in retrospect, why not push the new hash to the front of the stack? This would make your FETCH faster, so that it can return as soon as a key-value pair is found. Modify code as needed.

japhy -- Perl and Regex Hacker