Newsgroups: comp.compression
Path: utzoo!utgpu!cunews!hobbit.gandalf.ca!dcarr
From: dcarr@hobbit.gandalf.ca (Dave Carr)
Subject: Re: Greedy weirdness...
Message-ID: <1991Jun11.141044.5052@hobbit.gandalf.ca>
Organization: Gandalf Data Ltd.
References: <1991Jun10.230518.29430@mathrt0.math.chalmers.se>
Date: Tue, 11 Jun 1991 14:10:44 GMT
Lines: 13

In <1991Jun10.230518.29430@mathrt0.math.chalmers.se> d5kwedb@dtek.chalmers.se (Kristian Wedberg) writes:

>Not to be outdone, I've invented my very own dictionary based compression
>.... What I don't understand is why there is a optimal maximum for the 
>stringlengths.

It would depend on your update model.  It standard LZW, one character
at a time, then I am puzzled too.  Long strings could only get in the
dictionary if the string minus 1 character got used.

If, however, your update was done by concatenating 2 strings (ala 
V.42 bis), then the longer strings may never get used,
but they are taking up space in the dictionary.