Re: Reply on Burg and Welch power spectra

[Paul Kienzle]

On Sep 25, 2006, at 11:28 AM, Peter Vernon Lanspeary wrote:

> > Date: Thu, 15 Jun 2006 18:43:00 +0200
> > > 1) welch_psd.m   -- (Welch 1967) averaged-periodogram method
> > > 2) burg_filter.m -- Burg (1968) lattice-filter algorithm
> > > 3) mem_psd.m     -- spectral estimation from Burg coefficients
> >
> > Can you please compare your implementation of the pwelch to the
> > octave-forge version and to Matlab 2006a for compatiability? What 
> > would
> > it take to create a fully Matlab compatiable pburg function? Once that
> > is done these should be included in the signal processing toolbox of
> > octave-forge...

The interfaces were compatible when I created them in 1999.
The Mathworks has since made incompatible changes.  Given
that the main reason for compatibility is so that we can
run code available on the net, it hurts to break perfectly
good functions.

Would it be possible to allow individual package directories
to have their own path definitions?  That way old packages
could explicitly link to old versions.

Probably too much trouble for this case...

- Paul