timezone: long vs char*

[Erik Trulsson]

On Mon, Nov 09, 2009 at 11:31:06AM -0500, Garrett Wollman wrote:
> <<On Sun, 08 Nov 2009 19:41:03 -0500, "Joseph S. Atkinson" <jsa.bsd at gmail.com> said:
> 
> > Hello. I was working on bringing audio/liboggz up to date in ports, but hit a 
> > snag. liboggz expects timezone information to be long, which is (I am told) in 
> > accordance with SUSv3. FreeBSD is appears to be returning char*.
> 
> > I asked about this in #bsdports and was told that it appears that FreeBSD is 
> > wrong in this case and should be fixed. I saw no mention of this as a known 
> > issue on the standards webpage.
> 
> I think the people on "#bsdports" (an IRC channel, I assume) are
> wrong.  FreeBSD has never targeted conformance to any version of SUS,
> which memorializes a large number of UNIX System V's historic
> mistakes.  We do target conformance to POSIX (IEEE Std. 1003.1-2008 is
> the current version).

Well, these days the SUS and POSIX standards are pretty much the same thing.


> 
> In the POSIX specification, the declarations of "timezone" and
> "daylight" in <time.h> are flagged as "XSI", meaning that they are
> part of the X/Open System Interfaces extension, which in turn means
> that they are part of SUS but not POSIX.

If the POSIX specification specifies them they are very much part of POSIX,
but in this particular case an optional part of POSIX.


-- 
<Insert your favourite quote here.>
Erik Trulsson
ertr1013 at student.uu.se