Re: CGI, semicolons, and so on...

robm@ncsa.uiuc.edu (Rob McCool)
Message-id: <9312302002.AA27667@void.ncsa.uiuc.edu>
From: robm@ncsa.uiuc.edu (Rob McCool)
Date: Thu, 30 Dec 1993 14:02:32 -0600
In-Reply-To: Fisher Mark <FisherM@is3.indy.tce.com>
       "Re: CGI, semicolons, and so on..." (Dec 30,  8:20am)
X-Mailer: Mail User's Shell (7.2.5 10/14/92)
To: tcemail!FisherM@dxmint.cern.ch, www-talk <www-talk@www0.cern.ch>
Subject: Re: CGI, semicolons, and so on...
Content-Length: 1136
/*
 * Re: CGI, semicolons, and so on...  by Fisher Mark (FisherM@is3.indy.tce.com)
 *    written on Dec 30,  8:20am.
 *
 * 
 * The opacity or virtualization of the present scheme is nice, as the user of 
 * the URL does not have to concern her or himself with whether it represents 
 * a physical file or it is data created on the fly -- he or she just uses it.
 
I agree.
 
 * As far as the overhead for a single stat() call, I think that it is pretty 
 * small in the scheme of things.  Why even try writing a server in Perl, an 
 * interpreted language (which I love), if you need to worry about the overhead 
 * of a single stat() call?  Why even write one in C or C++ rather than 
 * assembler if a single stat call() can make the difference between a server 
 * with good response time vs. a server with poor response time?
 */

Well, if it was one stat(), you would be right... however, at times it can
get up to five or six. The overhead for this isn't really significant on a
stock filesystem, but over NFS or AFS (especially AFS) it becomes
significant, and in some cases it becomes large enough to become prohibitive.

--Rob