Contradictory definition of integers #376

Infinity is described as real but not rational. However, R7RS defines (integer? obj) on inexact reals to be true iff (= x (round x)). Since (round +inf.0) => +inf.0, (integer? +inf.0) => #t. This contradicts the example (rational? -inf.0) => #f and the fact that integers are a subset of the rationals.

This should be changed to the R6RS definition in terms of the denominator, which does not have this issue.

I'd argue that (integer? +inf.0) => #f is implied already due to integers being rational, as you said. Since the infinity isn't rational, it can't possibly be an integer, even if (= x (round x)).

The spec language is unclear though, so I agree that it should be changed a bit. Maybe add an additional clause that the inexact real must also be rational.

Agreed on both points. I'll make a.proposed R7RS-small erratum. At least Chibi will need fixing.