should complex forms look like Real + %i* Imag?
the program fullratsimp() has the same performance, so maybe the patch should be there rather than radcan. The original radcan was written before domain setting was implemented. The lisp program fr1 does the work of fullratsimp and is used by radcan.
integrate, asksign, part of answer doesn't use result
elliptic_e(1,1.23) --> lisp numeric error (introduces complex %i)
mystery, fixed by repairing maxima.init file to do right file_search ...
mystery, fixed by repairing maxima.init to include proper file_search_... locations.
integrate, csc, pquotient
integrate, atanh, facexp, facsum