There is a Matlab-based package called Chebfun that (among other things) computes polynomial approximations of functions. See here: http://www2.maths.ox.ac.uk/chebfun/. I don't know anything about signal processing, but it appears that this sort of approximation process is the foundation of some kinds of filter design. Apparently, the "Parks-McClellan" algorithm is the one typically used for this purpose. The PM algorithm just looks like a form of Remez algorithm, and the Chebfun code claims to be a "modern" version of this.
So, has anyone tried to use Chebfun in place of the older Remez or PM codes? If so, do you find that it's any better?