🚀 go-pugleaf

Submitted-by: steve@taumet.com (Steve Clamage)

rfg@netcom.com (Ronald F. Guilmette) writes:
>My first question is simply this.  Is the practice of avoiding definition
>of a va_list type in <stdio.h> strictly required by the ANSI C standard?
>My (naive?) believe is that this practice *is* required in order to avoid
>non-standard pollution of the user's namespace.

ANSI section 4.1.2.1 in conjunction with 4.8 and 4.9 makes it very clear
that <stdio.h> cannot define the identifier "va_list".  It is legal for
a user program to include <stdio.h> without including <stdarg.h>.  If
<stdarg.h> is not included, the identifier va_list is not reserved.
Hence hacks like __va_list which are a conforming way around the problem.

>My second question is also a simple one.  In what other cases are such
>"hidden built-in type hacks" required as a result of other requirements
>in the ANSI C standard (or in POSIX 1003.1-1990)?

I can't speak to the POSIX question.  Any identifier which is not
reserved according to 4.1.2.1 and which is implicitly referenced but
not defined by a Standard C header must have some sort of work-around.
(Only those identifiers not otherwise reserved may be defined in a
conforming Standard C header.)

Some types, like size_t, must be defined in more than one header.
This means that some workaround is needed for these types as well
to avoid multiple definitions.
-- 

Steve Clamage, TauMetric Corp, steve@taumet.com
Vice Chair, ANSI C++ Committee, X3J16


Volume-Number: Volume 29, Number 72