Design #656
eval: in which ring should the result be?
Description
Currently eval(f,[xval,yval]) wants xval and yval to be in the same ring as f; and the result is then naturally in that ring too.
It would be quite natural to have evaluation produce a value in the coeff ring (if xval and yval are in the coeff ring, or integers, or rationals).
Do we want this? Do we want both possibilities?
History
#1
Updated by John Abbott about 9 years ago
Updated by Currently eval is written in CoCoA-5, and just checks args before calling subst.
I find the version which returns a result in the coeff ring to be more "natural". Unfortunately I cannot simply call LC(eval(...)) because that fails when the result is 0 (and it is less readable than straight eval(...)).
We could make eval accept evaluation coords either in the poly ring or in the coeff ring; and the result would then be in the same ring as the coords. What to do if all the coords are INT or RAT? I suggest putting the result in the coeff ring. This would be backward compatible, but it is a nuisance to document (and might be confusing for users?).
#2
Updated by John Abbott about 9 years ago
We could make obsolescent the version which wants coords in the poly ring -- so it still works but produces a warning message. I wonder how much code would be affected by that.
Or we could have two fns with similar names: one for coords and result in coeff ring, and the other which works as the current fn.
What do you think?