char-set-contains? and #<eof>

char-set-contains? and #<eof>

[No Itisnt]

It's getting pretty annoying that char-set-contains? rejects #<eof>.
Is there a way around this without wrapping it myself? Is that
behavior intended? Because to me, it seems like it'd be more
reasonable to return #f when passed #<eof>.

Re: char-set-contains? and #<eof>

[Ludovic Courtès]

Hi,

No Itisnt <theseaisinhere@gmail.com> writes:

> It's getting pretty annoying that char-set-contains? rejects #<eof>.

It makes sense to me given that SRFI-14 (informally) defines the second
argument as being a character (and #<eof> is not a character):

  http://srfi.schemers.org/srfi-14/srfi-14.html#char-set-contains-p

Thanks,
Ludo’.

Re: char-set-contains? and #<eof>

[No Itisnt]

R5RS specifies that character predicates should return #f, and Guile's
character predicates are implemented in terms of char-set-contains?
R6RS says the same for char-set-contains?

The reason it bothers me is because it makes it cumbersome to deal
with input from a port. Case in point:

> $ grep "(and (not (eof-object\?" -R module
> module/sxml/upstream/SSAX.scm:		 (and (not (eof-object? c)) (char-alphabetic? c) c)) port)))
> module/language/elisp/lexer.scm:  (and (not (eof-object? tested))
> module/language/ecmascript/tokenize.scm:  (and (not (eof-object? c))
> module/language/ecmascript/tokenize.scm:              (cond ((and (not (eof-object? c)) (char-numeric? c))
> module/language/ecmascript/tokenize.scm:            (cond ((and (not (eof-object? c)) (char-numeric? c))
> module/system/repl/repl.scm:        (if (and (not (eof-object? ch)) (char-whitespace? ch))

Re: char-set-contains? and #<eof>

[Andy Wingo]

Hi,

On Fri 28 May 2010 22:52, No Itisnt <theseaisinhere@gmail.com> writes:

> R5RS specifies that character predicates should return #f,

One may interpret the language of the R5RS to permit raising an error if
the argument is not a character. Same with the R6RS. This is the tack
that Guile has taken.

For example, what would you have Guile do here:

  (even? #f) => ?
  (char-whitespace? 42) => ?

> The reason it bothers me is because it makes it cumbersome to deal
> with input from a port.

True; but it's just how it is, I think. The EOF object is not a
character.

Regards,

Andy
-- 
http://wingolog.org/