Re: A good hash algorithm

[Luiz Henrique de Figueiredo <lhf@...>]

>Is being satisfied with its "performance" sufficient?  You should still be concerned with minimizing collisions too shouldn't you?

I meant that we're satisfied with its overall performance, and I did not mean
just time, I meant collisions too, but I don't have any hard data on this
right now. We did do several collision tests, including looking at how *many*
keys were spread by the hashing algorithms (we once collected *all* identifiers
in *all* C programs available here at Tecgraf and the spread quite well).

Anyway, thanks for letting us know about Bob Jenkin's site.
--lhf