From bug-request at octave dot org Wed Apr  5 14:48:45 2006
Subject: Re: system() fails in high memory situation
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: "Tom Holroyd (NIH/NIMH) [E]" <tomh at kurage dot nimh dot nih dot gov>
Cc: Paul Koufalas <pkoufalas at adam dot com dot au>, bug@octave.org
Date: Wed, 5 Apr 2006 15:48:25 -0400

On  5-Apr-2006, Tom Holroyd (NIH/NIMH) [E] wrote:

| Perhaps system() could be changed to be more robust in this case,
| yes, and I might mention that to the Fedora people, but it's not so
| unreasonable to do something like the above in Octave, since large
| workspaces are common and modern boxes often have configs like mine,
| with 2 GB of stuff (mostly in swap) and the rest of the system is
| still working fine.  However, I haven't investigated the other paths
| that Octave might use to fork/exec.

I think the right thing for Octave to do in this case is call system()
and assume that it is implemented correctly.  If there is a problem
with that, then the correct place to fix it is in the implementation
of the C library system() function, not Octave.  Fixing it in Octave
would only help Octave.  Fixing it in the C library would help all
programs.

Or, perhaps there is a reason that it can't be fixed in the C library
because of some standards conformance issue or something else that we
haven't thought about.  In that case, it would not surprise me if the
same issues would affect Octave, and it would still be best to
stick with the C library function for most users.  In any case, if you
need this change for your specific application, then you have the
source, so you can make the change for yourself.

jwe



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------