RFC: Applet chrt, _POSIX_PRIORITY_SCHEDULING

[Harald Becker]

Hi Rich !

On 09-07-2013 01:43 Rich Felker <dalias at aerifal.cx> wrote:
>On Tue, Jul 09, 2013 at 07:16:51AM +0200, Harald Becker wrote:
>> Hi Bernhard !
>> 
>> >> > Bugzilla Bug #14829 decribe the underlying problem.
>> >> > http://sourceware.org/bugzilla/show_bug.cgi?id=14829
>> >> Note that, as discussed in the glibc bug tracker, glibc may
>> >> in the future remove the definition of
>> >> _POSIX_PRIORITY_SCHEDULING since Linux cannot satisfy its
>> >> requirements. ...
>> >
>> >Not exactly news, guess why I put this warning in..
>> >
>> >I'll remove the warning tomorrow.
>> 
>> Having a warning is not wrong. IMO we should have such a
>> warning if _POSIX_PRIORITY_SCHEDULING is missing, but it
>> should be a more informational message, like:
>> 
>> "process scheduling may not behave in conformance with POSIX"
>
>The amusing part is that the chrt command is probably
>interesting ONLY in the case where the sched_* functions are
>non-conforming. The POSIX versions are virtual no-ops on systems
>without the process contention scope (all they do is save and
>read back some data addressed by pid) and nearly no-ops even
>with the process contention scope.
>
>As such, I don't think the warning here makes sense...

At least a describing comment in the source shall give
information of the problem, else the next one steps on the same
topic sooner or later, and produce unnecessary noise on mail
lists / IRC channels :(

--
Harald