From bug-octave-request at bevo dot che dot wisc dot edu  Mon Feb  2 16:27:33 1998
Subject: Unidentified subject!
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: christian dot buchta at wu-wien dot ac dot at
Cc: bug-octave at bevo dot che dot wisc dot edu
Date: Mon, 2 Feb 1998 16:27:17 -0600

On  2-Feb-1998, Christian Buchta <buchta at eternity dot wu-wien dot ac dot at> wrote:

| To: bug-octave at bevo dot che dot wisc dot edu
| Subject: problem with built-in variable e
| 
| Bug report for Octave 2.0.9 configured for i386-suse-linux-gnu
| 
| Description:
| -----------
| 
|   First of all, thanks to the people who have developped this
|   software and made it available for everyone, it's really
|   useful for my scientific work.
| 
|   I think the best way to see what the problem is, is that I
|   copy the error-producing script and m-file into the next 
|   section. The first should have name bug_script.m and the 
|   second am_corr.m

This should be fixed in 2.0.10.

The fix was to install all built-in `constants' (like i, j, e, pi,
etc.) as functions in the symbol table (doing this simply changes the
semantics of these symbols -- the overhead of a function call is not
actually required to evaluate them).  The `constants' i and j have
always been installed as functions, but the others were not.  I don't
remember why i and j alone were special and I can't think of any
reason that e or pi should be any different from i or j.

The semantics for built-in variables like `do_fortran_indexing' have
not been changed, so modifying them within a function will change the
global symbol value.

Thanks,

jwe