Re: I want to contribute (give) two functions to the signal package

[Mike Miller]

On Wed, Jul 24, 2013 at 14:03:34 -0400, James-Adam Renquinha Henri wrote:
> I made two functions that allow the design of Legendre-Papoulis filters (see
> http://www.crbond.com/papers/lopt.pdf). These are
>
> function [z,p,k] = legendap(N)
> function [a,b,c,d] = legendf(N, Wp, varargin)
>
> The first gives the zeros, poles and gain of a analog lowpass prototype
> filter, and the second one acts like the butter function but generating
> Legendre-Papoulis filters; in fact I literally copied butter.m and edited it
> to use the legendap function.

Sounds good to me.

> Known issues:
> - The algorithm to generate the poles in legendap is not numerically stable
> and efficient; I proceed exactly like in the above pdf and there is an
> inevitable loss of precision when finding the roots of the polynomial. Using
> orders of 30 or above yields erratic but stable filters, and that may happen
> for orders below 30 but I didn't verify this.

Is this something you would want to look at addressing automatically
in the function or mentioning in the documentation?

> - legendap and legendf would require the appropriate test script in the
> footer.

Is that something you can add as part of your contribution?

Have you documented the functions using Texinfo format?

> The functions would be released using the GPL licence.

The signal package is licensed GPLv3, is that ok?

> I want to know if such a contribution is welcome and where would I "svn
> commit" that code.

Yes, your contribution is welcome. If you are familiar with mercurial,
you can fork the repository from

  http://hg.code.sf.net/p/octave/signal

and send a pull request to pull changesets from your clone hosted
somewhere on a public server. Or you can simply submit files to the
Octave patch tracker at

  https://savannah.gnu.org/patch/?func=additem&group=octave

Thanks for your interest in helping to improve the signal package!

-- 
mike