>>>>> "Richard" == Richard Fateman <fateman@cs.berkeley.edu> writes:
Richard> Depends on what you mean by "added functionality". If you add a "new
Richard> representation"
Richard> then I think the answer is yes, you have to consider what each command
Richard> might do, and whether
Richard> the default for "unknown" stuff is good enough. Added functionality
Richard> which fits in the standard
Richard> representation is not so problematical. E.g. adding the Lambert W
Richard> function which would require
Richard> a handful of additions:
Richard> a. how to differentiate
Richard> b. how to evaluate in floating point
Richard> c. how to evaluate in bigfloat
Richard> d. maybe some integration formulas
Richard> e. some fact about the inverse (for solve)
Richard> f. maybe some facts about taylor series
Richard> g. simplification at special arguments.
Let me also add item h. Limits.
Limits are particularly hard because you have to dig into the routines
and patch in some appropriate stuff. I regret not taking any notes or
adding documentation when I added some limits for the elliptic
functions. It took me some time to figure that out (probably because
I was stupid).
I have some preliminary stuff in maxima to evaluate some integrals to
get elliptic functions, but I don't even know where to begin to
integrate that with the integrate function.
Ray