On 04/16/2018 03:15 PM, Waldek Hebisch wrote: > AFAICS several patches make code more complicated, I am not > shure if we want such changes. In particular, while in > Factored exponents should be nonnegative it make perfect > mathematical sense to also have negative exponents. > To make things more clean we could have a separete domain > allowing negative exponents and Factored could inherit > from this domain most of its operations. That just a thought, > but it seems that disallowing negative exponents in > implementations leads to more complicated code.

Logically, Qian is right, but I often avoid NNI or PI, because then I would have to write qcoerce(n::Integer - 1)@PI just to get the types right while working in PI and knowing that n-1 will be positive. I would support Qian, if a-b inside PI or NNI could be written in a more natural way. (I still don't like SubDomain, since inheritance of the function signatures is not clear.) > Also changing recursion to loops seem to complicated code. Since we have restricted recursion depth and I've already had problems with deep recursion, I am in favour of loops-instead-recursion. Maybe, at some places it would be a good idea to hint in the comments at the recursive nature of the loop, if the recursive way looks much simpler. Ralf -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To post to this group, send email to fricas-devel@googlegroups.com. Visit this group at https://groups.google.com/group/fricas-devel. For more options, visit https://groups.google.com/d/optout.