Re: Use of _XOPEN_SOURCE?

[Corinna Vinschen <vinschen at redhat dot com>]

On Mar 17 11:38, Howland Craig D (Craig) wrote:
> Corinna wrote Wednesday, March 17, 2010 8:33 AM
> > Subject: Re: Use of _XOPEN_SOURCE?
> >
> > I don't know either, but, here's a question.  What about values
> between
> > 500 and 600 or between 600 and 700?  Shouldn't the code be more
> > forgiving, like this:
> >
> >   #if (_XOPEN_SOURCE - 0) >= 700        /* POSIX.1-2008 */
> >     #define _POSIX_C_SOURCE       200809L
> >    #elif (_XOPEN_SOURCE - 0) >= 600     /* POSIX.1-2001 or 2004 */
> >     #define _POSIX_C_SOURCE       200112L
> >    #elif (_XOPEN_SOURCE - 0) >= 500     /* POSIX.1-1995 */
> >     #define _POSIX_C_SOURCE       199506L
> >    #else
> >     #define _POSIX_C_SOURCE       2
> >   #endif
> >
> I purposely made it exact because as far as I read directly from the
> standards, POSIX requires those exact values for _XOPEN_SOURCE;  if
> another value is given for _XOPEN_SOURCE, there is not a pre-determined
> value of _POSIX_C_SOURCE to assign.  GLIBC does a < 600 thing, but
> they're taking a liberty not granted by POSIX.
>      My proposal does have one "forgiving" feature, put in for the older
> stuff before POSIX began the present method of exact values for
> _XOPEN_SOURCE and replaced the orignal fuzzy _POSIX_SOURCE with
> _POSIX_C_SOURCE:
>    #elif (_XOPEN_SOURCE - 0) < 500      /* really old */
>     #define _POSIX_C_SOURCE       2
>   #endif
> (The value of 2 does not have the same amount of specificity that the
> year/month values do.)
>      So, I still think that we should do it as I submitted.  If we
> were to do differently, to be a little more forgiving, I propose that
> it should purposely avoid values > 700.  If and when POSIX puts put
> a new release, we can add the proper value at that time.  (Presumably,
> they'll require 800 for their next release.)

Ok, I checked it in.


Thank you,
Corinna

-- 
Corinna Vinschen
Cygwin Project Co-Leader
Red Hat