> (C9) ratsimp(p/sol);
> (D9) (2*x^2+2*x+2)/(2*x+SQRT(3)*%I+1)
> With gcd == subres, the common factor of p and sol cancel.
> This doesn't happen with gcd == spmod. Is this a spmod bug?
> A subres bug? Or neutral? I get the same behavior with my macsyma.
With gcd:'algebraic, Maxima uses subres (called oldgcd internally) if
one or both arguments to gcd are algebraic (by testing whether kernels
have tellrat rules) *and* algebraic:true; otherwise, it uses spmod (zgcd
internally). So apparently it was a known issue when the code was
written.
It looks like you should use gcd:'algebraic instead of gcd:'spmod. In
fact, that might be the best default.
-s