Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!amdcad!sun!pitstop!sundc!seismo!uunet!mcvax!tuvie!tuhold!thom
From: thom@tuhold (Thom Fruehwirth)
Newsgroups: comp.lang.prolog
Subject: Re: Soundex revisited
Message-ID: <1178@tuhold>
Date: 26 Aug 88 13:43:32 GMT
Organization: Institut f. Angewandte Informatik, TU Vienna
Lines: 14

Commenting on my recent article, Richard O'Keefe <305@quintus.UUCP>
tried to proof by illustration that my implementation version of
soundex is too ugly.

Richard O'Keefes implementation is based on a different
specification, which is easier to implement. Second, his version
is only half-way between my specification and implementation version
To yield the same efficiency, a joining of his three predicates
called in soundex_chars/2 into one is necessary. Note that the
resulting version is arbitrarily faster than the current version
depending on the length of the word (how about 'CodesWithoutDuplicates')
one codes. The reason for this is that my  version
never codes more than three digits, while O'Keefes version codes
the whole word.