From: MikkelFJ <mikkelfj-anti-spam@...>
Date: 2003-10-10T01:13:54+09:00
Subject: Re: Is Ruby slower?


"Yukihiro Matsumoto" <matz@ruby-lang.org> wrote in message
news:1046421624.833733.22618.nullmailer@picachu.netlab.jp...
> Hi,
>
> In message "Re: Is Ruby slower?"
>     on 03/02/28, Travis Whitton <whitton@atlantic.net> writes:
>
> |All it would take would be converting it over to fit the variables in
st.c and
> |adding another preprocessor conditional such as #elif HASH_JENKINS so as
not to
> |disturb the 3 other strhash implementations. Anyway, just an idea. You
can read
> |about Bob Jenkins hashing function at
> |http://burtleburtle.net/bob/hash/doobs.html
>
> I will examine, when I have time to do.  Probably we need to remove
> prime modulo from st.c first.

I have been working with Jenkins hash - I can't say if is the best, but it
appears to be pretty good.
I did an optimization on the hash function to operate on exactly 32bits in
order get a fast integer map that grows and shrinks dynamically and which
remembers the order of insertion - so it is an indexed list. I'm not sure if
is any good in Ruby, but it is available upon request. It's only really
faster than red/black trees (STL) when there many elements, more than 1000
elements and possibly slightly slower with less than 100 elements.

Mikkel