AA with complex keytype?

[Frits van Bommel]

Manfred Nowak wrote:
> Frits van Bommel wrote
>> Your proposed default implementation doesn't satisfy that.
> In fact those requirements are necessary only for binary tress built in 
> the collision buckets.

Yes, but the current implementation *does* use binary trees, IIRC.

> If one can live with linear lists in the collsions baskets, then
> opCmp(Object) is allowed to degenerate to !opEquals(Object).

That might not be a great solution, especially when you need to consider 
that user-defined hash functions may not be very good...