>> Not guaranteeing they are fixnums [...]
>
> Saying they are exact integers in the range 0-127 *does* guarantee they are
> fixnums.

I mean that the user can pass in integers that overflow the fixnum
range. For example (ascii-upper-case? #x99999999999999999999). I'd like
an R6RS implementation of ascii-upper-case? and the others to be able to
use fx<? to compare the integer passed by the user.

> Okay, leave these uses of "fixnum" for offsets, then.

It's not just the offsets; see above.

I'm still slightly confused but I attempted some new language re:
fixnums and integers in the next draft; please let me know what you think.

>> I think the name
>> "ascii-horizontal-whitespace?" is far too long; hence it should remain
>> "space-or-tab?" or be abbreviated somehow.
>
> That's why we have code completion in Emacs and other IDEs.  But if it
> troubles you, you could drop "whitespace" from the name.  If an abstraction
> is worth a predicate, it's worth having a name for it. (Ditto with "ASCII
> codepoint".)

Lots of Scheme is written without code completion, it's still hard to
read longer identifiers and they cause long lines which forces people to
put more line breaks in code; and space-or-tab? is a very common and
simple operation. That's why I think we should limit the length of its name.

ascii-horz-white? would be OK with me but it's a bit weird since the
name of ascii-whitespace? is unabbreviated. And we need to have that
name because Scheme has had char-whitespace? since R2RS (!) so there's
no sense in gratuitously breaking from this time-tested convention.

space-or-tab also has the advantage that practically everyone
understands it instantly.