RE: hashing

["Steven Deller" <deller@smsail.com>]

> -----Original Message-----
> [mailto:comp.lang.ada-admin@ada.eu.org] On Behalf Of chris.danx
> Sent: Wednesday, April 10, 2002 5:31 PM
> To: comp.lang.ada@ada.eu.org
> Subject: hashing
> Hi,
> 
> Does anyone have a good hash function for strings (length is 
> arbitary, number of buckets is determined when package is 

I'd check out the PERL 5 sources.  The hashing function there seems to
be excellent for numerous cases I have looked into.  Not sure, but the
internal documentation might just explain the algorithm without having
to reverse engineer C code.

Other than that, I'd try one of the numerous ISO-defined hash functions,
such as CRC-32, etc.

I did a web search on "hash function algorithms" and came up with
NUMEROUS interesting sites.  

>...> 
> p.s. does anyone have any uses for circular lists?  Tom Swans 
> Tp5.5 book mentions them, what are they used for?  Just curious...

I have used circular lists in numerous buffering situations where a
simple double-buffer is insufficient.  Typically the situation is one
where the list is created once and the "head" and "tail" positions move
as data is put in and taken out. Using a circular list allows different
sizes to be put in and taken out.   I've also used a 10ary-buffer system
(the buffers themselves formed a circular list) to meet some system
transfer and error-correction requirements.  Some transmission systems
use a (short) counter of items sent and look for return ACK's for the
items; that algorithm is easy to write using a circular list.

> p.p.s. should performance/complexity information be included? 
>  I seem to recall a conversation on this a while back, but 
> can't remember if ppl thought it was a good idea.

Complexity information always raises its head at some point.  Best to
include it.

Regards,
Steve