Date: Mon, 10 Nov 1997 10:11:45 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Charles Sandmann <sandmann AT clio DOT rice DOT edu>
cc: sjmachin AT lexicon DOT net DOT au, djgpp-workers AT delorie DOT com
Subject: Re: malloc()
In-Reply-To: <9711091811.AA11422@clio.rice.edu>
Message-ID: <Pine.SUN.3.91.971110101111.13667C-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Sun, 9 Nov 1997, Charles Sandmann wrote:

> While this is true, I don't think emacs is represetative of all programs
> out there - it's just a huge application - and no more important than 
> any other program out there.  Using emacs as the test bed ends up with
> a malloc package optimized for - emacs.

Emacs is an example of a program that allocates, frees and reallocates
memory like heck.  Using system `malloc', it winds up using all
available free RAM and starts paging after several hours of serious
work.  That's why it comes with its own malloc package that overcomes
this problem.

While no single program is representative of all programs, and
therefore should not be the *only* program to be tested, Emacs will
surely make the sample more representative.  Different programs with
different memory-allocation patterns should be used, to make the
testing results more objective.