ellip is broken

[David Bateman]

Jan-Pieter Jacobs wrote:
> To: bug at octave.org
> Cc: jpjacobs
> Subject: ellip is broken
>
> Bug report for Octave 3.0.1 configured for i486-pc-linux-gnu
>
> Description:
> -----------
>
>   * The ellip command is broken apparently:
>   	This is what i get when executing the ellipdemo:
>
> 		[b,a]=ellip(5,1,90,[.1,.2]);  [ENTER]
> 		error: `fminbnd' undefined near line 37 column 4
> 		error: evaluating assignment expression near line 37, column 3
> 		error: called from `__ellip_ws' in file `/usr/share/octave/packages/signal-1.0.7/__ellip_ws.m'
> 		error: evaluating assignment expression near line 40, column 3
> 		error: called from `ncauer' in file `/usr/share/octave/packages/signal-1.0.7/ncauer.m'
> 		error: called from `ellip' in file `/usr/share/octave/packages/signal-1.0.7/ellip.m'
> 		error: near line 26 of file `/usr/share/octave/packages/signal-1.0.7/ellipdemo.m'
>
>   
The "ellip" function is part of the octave-forge signal package and not
the Octave core. Therefore this bug report should have been sent to
octave-dev at lists.sourceforge.net.. However, most developers are on both
lists so no problems..

When I run the "ellip" function on 3.0.1 with the latest release of
ellip I get

octave:1> [b,a]=ellip(5,1,90,[.1,.2])octave:1> [b,a]=ellip(5,1,90,[.1,.2])
b =

 Columns 1 through 6:

   1.3213e-04  -6.6395e-04   1.4925e-03  -1.9624e-03   1.4425e-03  
9.3884e-19

 Columns 7 through 11:

  -1.4425e-03   1.9624e-03  -1.4925e-03   6.6395e-04  -1.3213e-04

a =

 Columns 1 through 6:

     1.00000    -8.64826    34.60318   -84.21550   137.92762  -158.75980

 Columns 7 through 11:

   130.04250   -74.86358    29.00444    -6.83593     0.74556

and the same under matlabR2007b with the signal processing toolbox
installed. So it seems to work for me. However looking at the way in
which it fails for you the fminbnd is not found. fminbnd is part of the
optim package of octave-forge, and so the signal package should depend
on the optim package and it appears that it doesn't. I just added this
dependency to the SVN of the signal package.

Now to the workaround for you.. Can you please run "pkg list", and check
for the signal and optim packages.. You'll see a "*" next to the signal
package and either you'll see the "optim" package without the "*" or you
won't find the optim package at all. In the first case you'll need to
run "pkg load optim" to load the optim package. However this loading
won't be permanent and the only manner in which to make it permanent
with the "pkg" function is during the packages installation. So to make
the autoloading permanent run

system(["touch ",fileparts(which("fminbnd")), "/packinfo/.autoload"])

within Octave. This maybe have to be run as root.

Now if the optim package is not installed you'll have to download it
from octave-forge and install it with

pkg install optim-1.0.3.tar.gz

The default is to install this package with autoloading, and so the
"-auto" flag to "pkg" is probably not necessary.

Regards
David

-- 
David Bateman                                David.Bateman at motorola.com
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary