Re: Importing FIXES functions from octave-forge, and a couple of other things

[Paul Kienzle]

On Apr 2, 2006, at 5:37 PM, David Bateman wrote:

> All of these patches should allow all but tf2zp.m and zp2tf.m to be
> deleted from the octave-forge FIXES directory if they are accepted. As
> for zp2tf.m and tf2zp.m I still don't know what bugs these versions are
> supposedly fixing.. The only thing I can see related to this is
>
> http://velveeta.che.wisc.edu/octave/lists/archive//bug-octave.2001/ 
> msg00171.html
>
> and the error in this message now doesn't appear to happen with the
> octave versions. So perhaps Paul you can confirm, but I think zp2tf.m
> and tf2zp.m can also be deleted. John, comments please and which  
> patches
> need more work before you'll accept them and what is that work?

David,

Thanks for taking on 'FIXES'.

The tf2zp on octave-forge produce wrong results for some systems
(if b has leading zeros the gain is incorrectly set). This is enough
to justify purging it, though with the caveat that it is much
faster and much easier to follow than that in Octave.  Also, the
matlab docs suggest that doing the conversion on the statespace
representation is more stable.  This is done in octave but not
octave-forge.

Given that octave tf2zp/zp2tf aren't failing on the cases I see in
the mailing list feel free to purge them.

E.g., the following were fixed before 2.1.69

    
http://velveeta.che.wisc.edu/octave/lists/archive//bug-octave.2004/ 
msg00071.html
   http://velveeta.che.wisc.edu/octave/lists/bug-octave/2002/287

- Paul