Message written at 19 Jan 1999 10:54:43 +0000
--- Copy of mail to kaeru@flash.net ---

pete wrote:
>           is it not possible to add another optional argument to oscili that
> will define the order of interpolation, with a default being 1?  i dont know
> if the math supports this, but it is an idea.

The maths could do it but I have a couple of problems with thsi idea.
First it is rather a lot of code, unless I precompute the formulae
upto some limit -- otherwise one woudl be solving linear equations as
part of the init.  Secondly I have been convinced that only odd orders
make sense as that gives some symmetry.  I have been pleased with the
cubic; I did compare a 32-entry table and oscil3 against a computed
sine way and the errors were extremely small.  As it is cubic is too
slow (it really needs a power-of-2 plus 2 guards, rather than 1 as at
present, if it is to regain some of that).  

I am aware that teh coding in table3 is not of the highest order, and
could be optimised.