[systemd-devel] [PATCH 0/3] move timeout handling out of the core

[Zbigniew Jędrzejewski-Szmek]

On Mon, Jun 30, 2014 at 12:33:18PM +0200, Lennart Poettering wrote:
> On Sat, 28.06.14 06:00, Zbigniew Jędrzejewski-Szmek (zbyszek at in.waw.pl) wrote:
> 
> Heya,
> 
> > this patch series continues the trend of moving stuff out of
> > the core into generators. As a prerequisite for 3/3, it was
> > necessary to move some code from core to shared. If comments
> > are not counted, series is line-count-wise neutral, but I think
> > it makes things cleaner and easier to introspect.
> 
> I think I agree with this. The only reason I could think of why this
> might not be the right thing to do is to be nice to folks who wrote
> native mount units who want to use the mount option. Because this way
> the mount option becomes specific to fstab and cannot be used
> otherwise... (Which actually is exactly what we do for
> x-systemd.automount, which makes this all feel quite OK)
Good catch. I didn't think about that. Maybe a middle ground of issuing
a warning? I'll add some code which complains when the option is seen
in a unit file.

> I never liked the patching of the device unit's timeout the way we so
> far did. I feel quite a bit more comfortable with the way you propose as
> we never patch, we always have readable files in the fs...
> 
> Anyway, looks good to me. We need to document though that both
> x-sytemd.automount and x-systemd.device-timeout= strictly only work in
> fstab and not otherwise.
OK.

Zbyszek