Is integration and differentiation code sepparate?

Except for endless versions of bugs such as

 (%i1) integrate(x^(1/(1+%i) - (1/2-%i/2)-1),x,0,1);
 Maxima encountered a Lisp error:  Error in PROGN [or a callee]: Zero divisor.

I think you'll need to move away from testing integrals of sums of powers
to more general kinds of integrands.

I'd guess that if you worked at this long enough, you could collect some new Maxima bugs. 
Random testing will most likely generate endless variations of the same bug (an undetected 
pole in a definite integral, for example). 

If idle electrons bother you, generating random tests might be amusing.

--Barton