Path: utzoo!attcan!uunet!decwrl!ucbvax!ucdavis!csusac!jetfuel
From: jetfuel@csusac.csus.edu (Dave Jenks)
Newsgroups: comp.unix.internals
Subject: Re: Need efficient way to read file in reverse
Summary: You're trying too hard.
Message-ID: <1990Sep19.055846.14004@csusac.csus.edu>
Date: 19 Sep 90 05:58:46 GMT
References: <1990Sep18.152818.1303@phri.nyu.edu>
Reply-To: jetfuel@csusac.UUCP (Dave Jenks)
Organization: California State University, Sacramento
Lines: 17

In article <1990Sep18.152818.1303@phri.nyu.edu> roy@alanine.phri.nyu.edu (Roy Smith) writes:
>	Is there a standard, portable, efficient way to read a file in
>reverse?  I'm doing random seeks in a file and occassionally want to be able
>to find the beginning of the line into the middle of which I just seeked.
>Right now, I'm using a simple-minded revgetc() I wrote which is basically:

I had to do the very same thing a number of years ago.  The most
efficient method was to read using read(2) for BLKSIZ bytes (#defined
in <stdio.h> or <stand.h> or someplace like that) and search the
buffer manually.  With a simple searching algorithm, this can be very
fast - even with a poor algorithm, it'll still be much faster than
asking for i/o for each character.  It's even portable.
-- 
=======================================================================
	    "Pro is to con, as progress is to Congress..."
                                                                      
                                           ...!uunet \