maxintfloat should return an ::Integer

[tpapp]

Contrary to its documentation which says (emphasis mine)

... maxintfloat returns the smallest positive integer ...

it actually returns floating point values.

I don't know if this is a problem with wording or the implementation. IMO returning integers would be more useful as one could perform exact arithmetic & comparisons with them around the edge of the representable region.

[StefanKarpinski]

The word "integer" is a bit ambiguous in this case it means "integer-valued floating point number", so we could replace the word with that in the docs. Changing the type returned would be a breaking change, so we're not going to do that but you can easily just do Int(maxintfloat()) to get an integer.

[tpapp]

Not quite, eg maxintfloat(BigFloat), but Integer(maxintfloat(...)) seems to do the trick.

I will make a PR about the documentation then, and perhaps mention this?

[StefanKarpinski]

Yes, that would be great!

[JeffBezanson]

I think it makes sense to return special values of a type as that type itself. Improving the docs sounds like the right thing here.