[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Scheme-reports] 7.1.1 lexical structure

Andy Wingo scripsit:

> WHITESPACE is specified as SPACE OR NEWLINE.  Should TAB be included?
> LF?  Some unicode category?

I don't think so.  Schemers don't use hard tabs anyway, and the other
Unicode whitespace characters are for special purposes only.  Supporting
them means the lowest-level character dispatcher has to be UTF-8 aware
on UTF-8 systems, whereas all other significant characters come from the
ASCII repertoire.

> What is the deal with PECULIAR IDENTIFIER?  Is +.+ useful for someone?
> It seems an odd production, given that implementations are free to
> extend the set of valid identifiers.  The R5RS was clearer here.

This allows identifiers like +foo+, that can't be mistaken for numbers.

> I am bothered by the SYNTACTIC KEYWORD section.  Whether or not
> something is a keyword depends on scope; only sometimes does it only
> depend on name.

Yes.  I'm wondering if all that syntax-dependent stuff, which dates back
to R4RS, shoudn't be removed.  Editorial ticket filed.

> Do people really use the #-is-a-placeholder-digit thing?  Yuk!  This
> should be allowed (as any implementation extension would be) but not
> required.

I tried to get rid of that, but the WG went for backward compatibility.

> The INFINITY -> +nan.0 seems a bit sloppy, naming-wise.

If you knows of a better 'ole, go there.  I tried for a bit to find a
better name.

> Also, DEFINITION -> (begin DEFINITION*); this does not allow
>   (begin (begin (define a 1) a))

Editorial ticket filed.

John Cowan   cowan@x  http://www.ccil.org/~cowan
Most languages are dramatically underdescribed, and at least one is
dramatically overdescribed.  Still other languages are simultaneously
overdescribed and underdescribed.  Welsh pertains to the third category.
        --Alan King

Scheme-reports mailing list