Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!cs.utexas.edu!sun-barr!texsun!playroom!cliffs
From: cliffs@sunrock.east.sun.com (Clifford C. Skolnick)
Newsgroups: comp.lang.c
Subject: Re: search
Message-ID: 
Date: 18 Aug 89 23:15:42 GMT
References: <184@stsim.UUCP>
Sender: news@playroom.UUCP
Lines: 21
In-reply-to: glenn@stsim.UUCP's message of 17 Aug 89 13:19:22 GMT

In article <184@stsim.UUCP> glenn@stsim.UUCP (Glenn Ford) writes:

   I am writing a small project (not related to work, personal) that has to
   do search's on a sorted database.  I search off the keyword which is string
   of N length and is unique to all other keys.  For right now I do a binary
   search off the database, but this tends to be slow when run off a floppy
   drive (yes, it has to run on a floppy drive in SOME cases).  My question
   is: Is there a faster, yet fairly easy to implement, routine that can
   search my database?  Faster than binary search..  I don't want B-Tree's.  My
   database is allready built.

   Glenn Ford
   ..uunet!ocsmd!stsim!glenn

Perhaps you could make a better guest on your binary search start.  Say
look at the first character and make a guess about where to start in the
file.  You can also branch on much less, less, equal, greater, and much
greater.  This will be a bit faster than a binary search.  You say you
are using a binary search now, so not much more logic is needed.

				Cliff