Thanks. Ray's hypothesis makes sense. There are macro-definitions that are
apparently not available in ordinary GCL maxima run-time that are needed for
compiling. It also explains why reading simp.lisp in, uncompiled, doesn't
run the tests -- maybe it is calling some functions named the same as the
macros, and the functions are wrong.
This, in combination with the recommendation to run the "make" twice,
suggests that the process for making maxima might benefit from some analysis
and repair.
RJF
>