Status as of Octave version 2.0.16/2.1.31

Interface changes
=================
lin2mu
   Octave tries to guess based on the input range whether to use 8 bit
   or 16 bit samples.  This behaviour is wrong for quiet 16 bit recordings.
   A warning is issued if Octave guesses it is receiving 8 bit input.

mu2lin
   Octave defaults to returning 8 bit samples.  I have left this behaviour
   in place even though it is a silly default.  Change the default to 0 for
   compatibility with matlab.

deblank
   Trims \0 as well as " ".

Interface extensions
====================
cross		(warn if receiving mixed row/column vectors)
contour		(default number of contours)
fftfilt		(accepts matrices)
freqz		(accepts frequency vector; plots filter response)
imagesc		(only display if not returning scaled image)
polyder		(accepts a*b, a/b; auto-reduction of common terms)
polyderiv	(accepts a*b, a/b; auto-reduction of common terms)
polygcd		(new function for auto-reduction on polyder)
rand/randn	(accepts "state" argument; use the Mersenne twister,
		 a long period random number generator; faster)
mu2lin		(accepts -1:1 audio range; accepts matrix; 3x faster)
lin2mu		(accepts -1:1 audio range; accepts matrix)
cross		(accepts matrices)
grid		(control x/y grids and grid density independently)
fsolve		(allow parameters through fsolve to user function)

Bug fixes
=========
tf2zp		(search site:www.octave.org for test case)
zp2tf		(search site:www.octave.org for test case)

Speedups
========
findstr
hankel
hilb
index
invhilb
kron
rindex
toeplitz
vander

