#517 is(x < inf) => true

[Stavros Macrakis]

Compare apparently assumes that everything is smaller
than inf:

is(x < inf) => true ??
is(1/x < inf) => true ??
is(tan(x)^2 < inf) => true ??
is(und < inf) => error OK

These are reasonable questions to ask, I think. Even if
we consider that expressions' values range over
standard, finite reals (and not inf/minf, which we reserve
for return results), comparing to infinity has a reasonable
mathematical interpretation, namely: is the expression
bounded by a finite value. Clearly x is not.

On the other hand, x <= inf should always be true, even
for x:'UND (which currently causes an error).

But perhaps we need a global theory of all this stuff.

[Robert Dodier]

Logged In: YES
user_id=501686

Same behavior observed in 5.9.3cvs.