Re: [PATCH 8/9] autoconf: Check for timer_settime

On Wed, 2014-09-10 at 14:08 -0700, Junio C Hamano wrote:

Karsten Blees [off-list ref] writes:

quoted

While the timer extension (timer_settime) has graduated to mandatory in
the current POSIX spec, the monotonic clock extension is still optional
today (i.e. not necessarily supported even on newer Unices). In contrast
to this, the XSI extensions seem to be widely supported.

IMO the 'obsolescent' marker in the current POSIX spec is no reason to
spring into action (at least not yet). E.g. utimes (also in <sys/time.h>)
has been marked LEGACY in the 2004 version and is no longer LEGACY today.
Btw., we'd also have to find a replacement for gettimeofday and probably
a lot of other stuff...

Therefore I tend to agree with Hannes that we should stick with setitimer
and emulate it on systems that don't have it (as we do on Windows).

Sounds sensible to me.

Ya that makes sense. Was thinking too much about compatability code for
Linux kernel drivers, which is a different issue.

Let's go with re-implementing settimer in terms of timer_settime on the
one system we have information that doesn't support settimer.

I could spin that patch, but I think the original author should, since
he's the one who could actually test it.

Regards,
Jake

--
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html