A regression in Maxima 5.17.1 with limits and assumptions?

On 12/23/08, Barton Willis <willisb at unk.edu> wrote:

>  getting that to work is a puzzle -- does it require our favorite ev?

I don't think so --- I think a macro + buildq is just right.

simp_using (e, [a]) ::=
  buildq ([e, a], block ([c%], unwind_protect ((supcontext (c%), apply
(assume, a), e), killcontext (c%))));

macroexpand (simp_using (asksign (x), x > 0));
 => block([c%], unwind_protect((supcontext(c%),
           apply(assume, [x > 0]), asksign(x)), killcontext(c%)))

load (unwind_protect);
simp_using (asksign (x), x > 0);
 => pos

Probably buildq should substitute a gensym for the name of the context.
Maybe supcontext() (no args) should gensym the context name;
at present supcontext complains if not given the context name.

I've convinced myself that syntax to help distinguish arguments
which have different purposes is usually a good idea.
In this case something like:

simplify asksign (x) assuming x > 0;

or maybe

simplify (asksign (x) assuming x > 0);

I like the latter; the obvious parallel is simplify(asksign(x)) (no
assumptions).
Of course, that suggests there could be a function named "simplify" ....

FWIW

Robert