Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/17/84 chuqui version 1.9 3/12/85; site unisoft.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!lll-crg!dual!unisoft!fnf
From: fnf@unisoft.UUCP
Newsgroups: net.unix
Subject: Re: fgrep (isn't)
Message-ID: <507@unisoft.UUCP>
Date: Tue, 23-Jul-85 13:50:36 EDT
Article-I.D.: unisoft.507
Posted: Tue Jul 23 13:50:36 1985
Date-Received: Thu, 25-Jul-85 21:43:04 EDT
References: <495@unisoft.UUCP> <5785@utzoo.UUCP>
Reply-To: fnf@unisoft.UUCP (Fred Fish)
Organization: UniSoft Systems, Berkeley
Lines: 41

In article <5785@utzoo.UUCP> henry@utzoo.UUCP (Henry Spencer) writes:
>> Notice that plain old grep is the fastest of all, and fgrep is the slowest!
>
>You forgot to test egrep, which is well-known to be the fastest for the
     ^^^^^^^^^^^^^^^^^^^^	    ^^^^^^^^^^
>trivial case.  Where fgrep makes a difference is when you want to search
>for a number of strings simultaneously.  Grep can't do that at all, and 
>egrep hits pattern-complexity limits quickly.  Fgrep really screams along
>in this case...

Thanks for the tidbit of Unix folklore Henry.  But no, I didn't forget
to test egrep, here is an extract from my original posting:

		real	user	sys
		----	----	---

	grep	1.9	1.1	0.6
	bgrep	2.7	1.9	0.7
	egrep	3.5	2.6	0.7
	fgrep	9.6	8.8	0.7


It would be nice if the documentation wasn't so misleading.  From the
From the system V User's Manual grep(1):

	"Grep patterns are limited regular *expressions* in the style
	of ed(1)..."

	"Egrep patterns are full regular *expressions*..."

	"Fgrep patterns are fixed *strings*; it is fast and compact."

This implies that fgrep should be the fastest for plain old strings
even if there is only one (or none? :-).

-Fred

===========================================================================
Fred Fish    UniSoft Systems Inc, 739 Allston Way, Berkeley, CA  94710  USA
{ucbvax,decvax}!unisoft!fnf	(415) 644 1230 		TWX 11 910 366-2145
===========================================================================