You don't want a two-way hash, you want two hashes that are kept in sync:

    my( %one, %two );
    Tie::Hash::Complementary->marry( \%one, \%two );
    $one{foo} = 'bar';    # Also sets $two{bar}= 'foo'
    $two{baz} = 'bif';    # Also sets $one{bif}= 'baz'
    $two{bar} = 'oof';    # Does: $one{oof} = delete $one{foo}
    $one{oof} = 'baz';    # Dies; $two{baz} can't be both oof and bif
[download]

Of course you can't have overlap between the keys with a single two-way hash. It is a single hash. You found a module that does something somewhat like what you want, but also fundamentally different than what you want.

Tie::Hash::TwoWay does actually store the mapping in two different hashes internally, it just doesnt have any way to specify which direction to operate on, it tries the "primary" direction, and if the key doesnt exist, the operation is tried on the "secondary" direction.

In any case, at lunch today, i designed a tie that does what i need, i'm about to implement it. I'll post about it, and submit it to CPAN, hopefully by the end of the day!

$name{'555-1212'} = 'Jones';
$phone{'Jones'} = '555-1212';
[download]

--
[ e d @ h a l l e y . c c ]

Yeah, unfortunately, many modules are not very robust, in that they dont allow for much flexibility. But what can we expect for free! (we do get plenty).

I want to have the operations encapsulated, otherwise, i'll need to perform operations on both hashes every time i want to do anything to one of them. So, i think im going to design my own variation, where you can specify in which direction the "key" operation should be dealt with. The interface may look something like:

    tie my %hash, 'Tie::Hash::MyTwoWay';
    ...
    $hash->primary()->delete("foo");
    $hash->secondary()->delete("foo");
[download]

Here the 2 deletes perform VERY different operations, in that they delete from different directions.

By default, an operation will take place on "primary"
For enhancements, i'll probably add optional aliasing of "primary" and "secondary".