> From: Jean Delvare <jdelvare@xxxxxxx>
> What really surprises me though is that Dale seems to see the problem
> on Fedora, while I thought Jaromir had already committed his fix there,
> regardless of upstream acceptance.
It's possible that the patch has been put into Fedora 20 but not into
Fedora 19 (which is what I'm running). (*Not* Fedora 16 as I
mistakenly wrote, ugh.)
> As a summary, I agree that there is a problem, but I am not a big fan
> of using wrapper shell scripts to solve it, for it defeats the purpose
> of systemd.
That's true. On the other hand, there's no way in systemd to mark an
ExecStart action in a service as "execute this if and only if (some
condition)".
(There are Conditional qualifiers you can put on a unit, but they seem
to apply to the unit as a whole, and we always want the ExecStart of
/usr/bin/sensors to run. If the ExecStart lines were put into Bash,
we could use shell conditionals to handle it, but that is precisely
what the design wants to avoid. I suppose you could split the
modprobe into a separate until that is to be executed before
lm_sensors.service, and have the configuration-detector do whatever is
needed to tell systemd to ignore the unit if no modules need to be
loaded.)
> Unfortunately I have to admit that I did not take the time to
> actually implement my idea in openSUSE, so I still don't know for
> sure if it would work OK.
As James Morris said,
Finally, I'll offer eight opinions that computer scientists who venture into
the real world should consider:
...
3. Virtual << Real.
...
We have to go with whatever exists.
Dale
_______________________________________________
lm-sensors mailing list
lm-sensors@xxxxxxxxxxxxxx
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
