proposal concerning conflicts between nset and other stuff in Maxima



(1) User-level and simplifying functions in nset
should autoload. The location of 'flatten' doesn't
matter. Except for bogus input (flatten(x^y^z)), the 
two 'flatten' functions should work the same.  Since
the flatten function in nset is the smaller of the two, 
I think src/flatten.lisp should be expunged. If
we do that, rtestflatten could be deleted as well
(these tests are in rtestnset).

(2) At one time 'generic-autoload' didn't 
recognize 'fas,' 'fasl,' or 'x86f' as extensions
for complied lisp code.  See nset/maxima-init.lisp for
a proposed fix.

(3) In rtestnset.mac, we should change (load ("nset.lisp"), 0)$
to (load ("nset"), 0)$ That way, if a user compiles nset, the 
compiled version gets loaded. 

Barton