Xref: utzoo comp.theory:346 comp.misc:8215 comp.lang.misc:4119
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!saturn!xanthian
From: xanthian@saturn.ADS.COM (Metafont Consultant Account)
Newsgroups: comp.theory,comp.misc,comp.lang.misc
Subject: Re: hashing function for strings
Message-ID: <10940@saturn.ADS.COM>
Date: 18 Feb 90 01:17:02 GMT
References: <1990Feb15.211210.22950@max.sunysb.edu> <0ZqzBRm00WA1M0d5JZ@andrew.cmu.edu> <7416@ogicse.ogc.edu>
Followup-To: comp.theory
Organization: Advanced Decision Systems, Mt. View, CA (415) 960-7300
Lines: 15

In article <7416@ogicse.ogc.edu> daniels@ogicse.ogc.edu (Scott David Daniels) writes:

>the hardware word size was thirty-two bits, and (since % produced inconsistent
>results on our two target machines for full-sized words),

Yep.  It is utterly amazing how many modulus inplementations exist
where (-1) mod m isn't m-1.  This makes cellular automaton programs
add half a dozen extra instructions per loop to treat a rectanglar
array as a toroid.  I don't know who originated this bogosity, but it
would be nice if hardware and compiler folks would extirpate it once
and for all from all machines and all languages, so programs using mod
could port without stumbling over this implementation error.
--
xanthian@ads.com xanthian@well.sf.ca.us (Kent Paul Dolan)
Again, my opinions, not the account furnishers'.