> I took it to be part of the definition of these functions (other than
> ascii?) that it's an error to apply them to non-ASCII characters, but I
> see that's not actually true.  Perhaps it should be?  Consider that
> (number? 'foo) returns #f, but (even? 'foo) is an error.

If we keep the ASCII predicates so that they accept non-ASCII characters
and return #f for them, what's your opinion on having an
`ascii-non-control?` procedure? I know it's a bit un-Scheme-like to have
the negations but can't think of anything better.

This ought to be the last open question pertaining to SRFI 175; as we
settle this, I'll publish a draft and announce last call.