From bug-octave-request at bevo dot che dot wisc dot edu  Mon Aug  9 14:09:37 1999
Subject: Re: empty matrices in aritmetical operations
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: Mirek Kwasniak <mirek at zind dot ikem dot pwr dot wroc dot pl>
Cc: bug-octave at bevo dot che dot wisc dot edu
Date: Mon, 9 Aug 1999 14:08:50 -0500 (CDT)

On  9-Aug-1999, Mirek Kwasniak <mirek at zind dot ikem dot pwr dot wroc dot pl> wrote:

| On Mon, Aug 09, 1999 at 09:53:23AM -0500, John W. Eaton wrote:
| > On  9-Aug-1999, Kwasniak Miroslaw <kwasniak at mirek dot ikem dot pwr dot wroc dot pl> wrote:
| [...]
| > | octave:17> []+[1 1]
| > | error: operator +: nonconformant arguments (op1 is 0x0, op2 is 1x2)
| > | error: operator +: nonconformant arguments (op1 is 0x0, op2 is 1x2)
| > | error: evaluating assignment expression near line 17, column 3
| > | 
| > |   Matlab gives always [] as result.
| > 
| > Sorry, I don't think this is a bug in Octave, except that to be
| > consistent with what happens for
| 
| I see some code from net and mathworks that use statemets like
| 
|   for i=..
|     ...
|     c=[ c a*b ];
|     ...
|   end
| 
| where a is empty (and b not) for first iteration. :(

Well, Octave tries to do the right thing mathematically for these
kinds of operations (check the references in the Octave manual).  To
allow

  A * B

to work when the dimensions of A and B do not conform and neither is a
scalar does not make a lot of sense to me.  It should be an error.
My advice is to fix the buggy code instead of trying to `fix' the
behavior of Octave, which I think seems quite reasonable.

jwe



---------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.  To ensure
that development continues, see www.che.wisc.edu/octave/giftform.html
Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
---------------------------------------------------------------------