cvs commit: src/etc Makefile sensorsd.conf src/etc/defaults rc.conf src/etc/rc.d Makefile sensorsd src/lib/libc/gen sysctl.3 src/sbin/sysctl sysctl.8 sysctl.c src/share/man/man5 rc.conf.5 src/share/man/man9 Makefile sensor_attach.9 src/sys/conf f

[Constantine A. Murenin]

On 18/10/2007 03:25, Alexander Leidinger wrote:

> Quoting Poul-Henning Kamp <phk at phk.freebsd.dk> (from Wed, 17 Oct 2007  
> 20:24:14 +0000):
> 
>> In message <47162F40.8000203 at FreeBSD.org>, "Constantine A. Murenin" 
>> writes:
>>
>>> You didn't quote my other part of the reply about the ntpd/sensors.c
>>> example.
>>
>>
>> FreeBSD already supports an API for timekeeping events, complying
>> with the only relevant standard in this area: RFC 2783
> 
> 
> He didn't talk about using this for timekeeping stuff in FreeBSD. He  
> used this an an example what could be done for RAID.

I confirm.

> As you didn't asked for a specific action regarding this I try to do  
> it. Please correct me if I don't get it right:
> 
> Constantine, we have a standards compliant API for timekeeping events.  
> Any time related stuff should use this API. Please add a text and a  
> pointer, e.g., in the man-page and/or any other sensible place, which  
> tells that our RFC 2783 compliant API shall be used.

Sure, in fact, I don't think such documentation is even needed -- 
OpenNTPD, the primary daemon that benefits from the timedelta sensors, 
resides in the ports tree, so any timedelta drivers could be implemented 
as loadable kernel modules and also reside in the ports tree, to be used 
together with OpenNTPD.  I see no technical or architectural violations 
here, merely an extra alternative for the end users.

The sensors framework itself, on the other hand, obviously cannot reside 
in the ports tree.

C.