Re: "A diagram of C23 basic types"

scott@slp53.sl.home (Scott Lurndal) writes:

Keith Thompson <Keith.S.Thompson+u@gmail.com> writes:

scott@slp53.sl.home (Scott Lurndal) writes:

Tim Rentsch <tr.17687@z991.linuxsc.com> writes:

scott@slp53.sl.home (Scott Lurndal) writes:

bart <bc@freeuk.com> writes:

[...]

So it is not true that you need include stddef.h, nor obvious
that that is where NULL is defined, if you are used to having it
available indirectly.

>
Indeed, and it is well documented.
>
For example, in the POSIX description for the string functions
you'll find the following statement:
>
    [CX] Inclusion of the <string.h> header may also make visible
         all symbols from <stddef.h>.  [Option End]
>
This is true for a number of POSIX headers, include those you
enumerate above.
>
[CX] marks a POSIX extension to ISO C.

>
How strange.  I don't know why anyone would ever want either to
rely on or to take advantage of this property.

>
Some existing unix implementations at the time the standard was adopted
had that behavior and the committee was not willing to break existing
implementations.

>
You mean the POSIX standard, yes?  The C standard does not permit
<string.h> to include <stddef.h>.

>
Yes, and POSIX explictly marks it as an extension to the C standard.
>
So, if unix/linux system header files are posix compliant, they're
technically not completely compliant with the C standard, although
they will compile code that complies with the C standard.

An implementation can be both C compliant and POSIX compliant by
ignoring that extension.  POSIX doesn't *require* inclusion of
<string.h> to include <stddef.h>.

But an implementation that does do that does not conform to the C
standard, since it will reject the following valid C code:

    #include <string.h>
    int wchar_t = 0;

(Unless the diagnostic is a non-fatal warning, I suppose, or unless
<string.h> makes wchar_t visible only if _POSIX_C_SOURCE is defined,
but why bother?)

To be clear, I'm not suggesting that defining wchar_t for your own
purpose is a good idea.

--
Keith Thompson (The_Other_Keith) Keith.S.Thompson+u@gmail.com
void Void(void) { Void(); } /* The recursive call of the void */