From owner-freebsd-current  Thu Mar  2 17:45:37 2000
Delivered-To: freebsd-current@freebsd.org
Received: from bugg.strangled.net (c705742-a.htfdw1.ct.home.com [24.2.137.94])
	by hub.freebsd.org (Postfix) with ESMTP id 84B6B37BD7B
	for <current@freebsd.org>; Thu,  2 Mar 2000 17:45:32 -0800 (PST)
	(envelope-from bugg@bugg.strangled.net)
Received: (from bugg@localhost)
	by bugg.strangled.net (8.9.3/8.9.3) id UAA08639
	for current@freebsd.org; Thu, 2 Mar 2000 20:47:20 -0500 (EST)
	(envelope-from bugg)
Date: Thu, 2 Mar 2000 20:47:19 -0500
From: Dan Papasian <bugg@bugg.strangled.net>
To: current@freebsd.org
Subject: which(1), rewritten in C?
Message-ID: <20000302204719.A8563@moe.c705742-a.htfdw1.ct.home.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 1.0i
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

While this may sound crazy, I was tired of 'which' taking a long
time to complete on my 486 dx4/100 when it was under extereme
pressure, so I rewrote it in C :)

I don't think that it is worthwhile replacing this perl script
with one written in C, but if anyone is interested...

- The overhead of calling perl isn't noticable on most machines

- which(1) isn't exactly something that is executed often

- Many modern shells have which as a shell built-in.  
  OTOH, that doesn't mean people using /bin/sh should be
  penalized.

- I believe that C is easier to read than perl.  Of course,
  this is a religous argument of sorts, so I don't think
  that this is a reason to replace it :)
			BUT
  IMHO having which be written in C makes it a little bit more
  consistent with the rest of the source base, and a useful tool
  no longer depends on perl.
			OTOH
  You get a larger binary size with C, and have to compile it.

moe% /usr/bin/time ./which perl
/usr/bin/perl
        0.03 real         0.00 user         0.02 sys

moe% /usr/bin/time which perl
/usr/bin/perl
        0.13 real         0.08 user         0.03 sys

NOTE:
	This version of which has exactly the same behavior.
	Also, the above test was not performed when the box was
	under load.. and on slower machines/under load, the
	differences are of course, more noticable.
	You may all go ahead and call me crazy now.

...I've got the fear of posting the source, but what the heck,
getting nitpicked is good education :)

http://bugg.strangled.net/which.c

Any flames^Wthoughts?


-Dan Papasian
<bugg@bugg.strangled.net>


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message