From bug-octave-request at che dot utexas dot edu  Fri Dec 17 12:22:02 1993
Subject: octave-0.79 memory leak
From: kircher at neuro dot tuwien dot ac dot at
To: bug-octave at che dot utexas dot edu
Cc: zorman at neuro dot tuwien dot ac dot at (Werner Zorman)
Date: Fri, 17 Dec 1993 19:21:10 +0100 (MET)

Hello,

we apparently found a memory leak in octave-0.79. It manifests itself on
iterated function calls. This is the minimal setup to produce the
effect:

Consider the following file dummy.m:

  function dummy
  end

After starting octave, ps gives the following:

USER       PID %CPU %MEM SIZE  RSS TTY STAT START   TIME COMMAND
gk        7827  1.0  3.8 2362  584 pp2 S    21:57   0:00 octave

now we execute

octave:1> for j=1:10000 dummy; end

after that ps gives 

USER       PID %CPU %MEM SIZE  RSS TTY STAT START   TIME COMMAND
gk        7827 47.4 15.2 4022 2292 pp2 S    21:58   1:54 octave

once again

octave:2> for j=1:10000 dummy; end

USER       PID %CPU %MEM SIZE  RSS TTY STAT START   TIME COMMAND
gk        7827 48.0 26.1 5662 3932 pp2 S    21:57   3:48 octave


When we use built-in functions instead of m-files the effect is
weaker. In a similar session with dummy; substituted by sin(0); the
three ouputs of ps are 

USER       PID %CPU %MEM SIZE  RSS TTY STAT START   TIME COMMAND
gk        7878  8.8  3.8 2362  584 pp2 S    22:08   0:00 octave
gk        7878 39.7  6.1 2674  928 pp2 S    22:08   0:11 octave
gk        7878 44.1  8.2 2986 1240 pp2 S    22:08   0:23 octave


I am sorry for not having really tracked the problem down, but I think
you will be able to locate the error very much faster than we are...

Sincerely

Gerhard Kircher