You can convert your data via something like:

    my $devs= hashjoin( {
        version=>\%vers, module=>\%mod, memory=>\%mem } );

sub hashjoin {
    my( $in )= @_;
    my $out= {};
    for my $attrib (  keys %$in  ) {
        for my $item (  keys %{ $in->{$attrib} }  ) {
            $out->{$item}->{$attrib}= $in->{$attrib}->{$item};
        }
    }
    return $out;
}
[download]

I could factor that a little differently but I chose this way because I'm not assuming that all of your hashes actually do have the exact same keys and I didn't feel like checking that assumption.

Here is yet another case where I really want my "dereference an entire list at a time" operators. Then I could replace that middle loop with:

    $out->@{ keys $in->{$attrib}->% }-->{$attrib}=
        values $in->{$attrib}->%;
[download]

and it would make it easier for you to build your hashes this way from the beginning. (And, no, that last bit of code isn't valid Perl -- it is what I wish Perl supported: -->, ->%, ->@, ->@{}, ->@[], ->$, etc.)

I'd love to see an elegant method for generating the data in the format that merlyn suggested from the beginning, if each attribute is gathered from a different subroutine.

        - tye (but my friends call me "Tye")