From: bonzini@gnu.org   
      
   > Specifically, I've been working on string matching problems recently,   
   > and thus the classic Aho-Corasik (AC) and Boyer-Moore (BM) algorithms.   
   > In addition, I've known for a "long time" that the trick of working   
   > backwards ala BM could be applied to a AC style matcher to obtain the   
   > same potential sub-linear performance.   
      
   I think GNU fgrep is doing something very similar.   
      
   The source code is at   
   http://cvs.savannah.gnu.org/viewcvs/grep/src/kwset.c?rev=HEAD&ro   
   t=grep&view=markup   
   and it cites "A String Matching Algorithm Fast on the Average,"   
   Technical Report, IBM-Germany, Scientific Center Heidelberg,   
   Tiergartenstrasse 15, D-6900 Heidelberg, Germany.  This paper is   
   available through Springer.   
      
   Paolo   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)