1. When Maxima asks questions like "is g1023 positive negative or zero?"
then it is a bug. It is asking about generated symbols internal
to a problem. This kind of bug continues to exist, and I
have seen it in commercial macsyma, too. I don't know
of any systematic way of removing these problems.
2. Maple has had, maybe still has, a convention in which
variables with assumptions look different. Like x~ instead of x.
I think that is it, anyway.
So we could have an expression that has a mark on it, and
by asking "what assumptions are there?" get them revealed.
This could improve the display, but doesn't really affect
the need for internally computing with the assumptions
so that they are simplified.
We could attach the assumption in a variety of ways, using
hash tables, or attaching info to the flags in the expression.
e.g. a+b is now ((mplus simp) $a $b). it could be
((mplus simp ...assumptions...) $a $b).
This kind of encoding would be difficult without changing
the simplifier.
RJF