🚀 go-pugleaf

diamond@jit.dec.com (Norman Diamond) writes:
>In article <*}s-pkdc@twinsun.com> eggert@twinsun.com (Paul Eggert) writes:
>>ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe) writes:
>>>    Am I correct that ANSI header files should be independent of identifiers
>>>    available to the user, using instead e.g.
>>>	void *malloc(size_t);		/* no identifier, or */
>>>	void *malloc(size_t _size);	/* reserved identifier */
>>You're basically correct.  A nit: `_size' isn't reserved (see ANSI C 4.1.2.1).
>>`__size' is, so it would work here.
>
>Mr. O'Keefe's complaint is correct, and his proposed adjustment was correct
>though on thin ice [personal opinion only].  Mr. Eggert's nit is wrong.
>
>The standard headers must be included outside of any external declaration or
>definition (ANSI 4.1.2, page 97 lines 31-32; there are also more requirements).
>
>_size is reserved for use as an identifier with file scope in both the
>ordinary and tag name spaces (ANSI 4.1.2.1, page 98 lines 7-8 and 16-18).
>I think the standard's wording is very unclear on whether _size can be
>declared as a macro, but footnote 88 clarifies the intention, and an
>implementation COULD depend on _size as a reserved identifier at file scope.

ANSI 3.1.2 pg. 21 line 5 "Macro names and macro parameters are not considered
further here", so macro names aren't in the "identifier with file scope"
category, and aren't reserved by 4.1.2.1 lines 7-8.

Footnote 88 says "macro names matching any of the reserved identifier names
must not be defined if an associated header, if any, is included".
                          ^^^^^^^^^^
"_size" is not associated with any of the headers, so I don't think it's
covered by the footnote, either.

Your interpretation seems to say that the user cannot define any macros that
start with an "_", at least if any standard header is included.

>So it's pretty hard to argue that the standard would allow the user to define
>_size as a macro in the part of a file where standard headers can be included.

Hey, it's the weekend, and I'm feeling reckless.

>(However, if I were an implementor, I would use two underscores in all of
>this kind of implementation stuff [personal opinion only].)

Ditto. And, as a user, I never create identifiers that start with an _:
it's a much easier rule to figure out and remember.

Reserved identifiers are strange and striking!
-- 
David Tanguay       dat@Thinkage.on.ca  dat@Thinkage.com  uunet!thinkage!dat
Thinkage, Ltd.           Kitchener, Ontario, Canada          [43.40N 80.47W]