On Sunday 02 March 2003 09:07 pm, Mike Campbell wrote:
> > > Sorry if this contribution is "newbie-like" as I'm not an expert
> > > in any kind of mathematical/analytical field, but... Just wanted
> > > to mention that my Sedgewick algorithms book goes into a bit of
> > > detail on why prime numbers should generate better distributed
> > > hash values using the modulo operator than powers-of-two...
> >
> > Mathematically, that is true.  However, if you analyze the computational
> > aspects, the impact is not obvious, as you are decreasing collisions at
> > the cost of a more expensive hash generation algorithm.  I'm not saying
> > that it definitely wouldn't pay off, but the answer is not obvious, and
> > my belief is that there would be a small net loss.
>
> Beliefs are fine, but if this is going to make a big difference either way
> in ruby, wouldn't a *test* be warranted?

Certainly.  I am just saying don't _assume_ it will be faster, or slower for 
that matter.  I would like to see the results of any testing.

-- 
Seth Kurtzberg
M. I. S. Corp.
480-661-1849
seth / cql.com