Re: Bug#676817: systemd and dovecot
Thanks for your responses. Based upon them I wail upload an NMU with a
2 day delay doing nothing more than limiting the systemd changes to
linux. I would not normally feel comfortable with such a short delay but
in this case ny proposed changes should be self-evident and uncontroversial.
On 17/06/12 11:26, Tollef Fog Heen wrote:
> ]] Nicholas Bamber
> I would be most grateful if you quoted the way is usually done on email
>> I would be very grateful if you could have a look. Andreas Barth has
>> basically repeated the point I made in the third paragraph of my
>> original post.
> Yes, and you're both mistaken. systemd is not a normal daemon package,
> it does not start any daemons, nor replace init merely by being
> installed. Installing systemd onto a system is about as intrusive to
> the system as a whole as installing nvi.
>> Nothing you have said is really reassuring me. You talk about how a
>> package needs something to make socket activation to work and
>> sd-daemon.h is a way to do that. Well that file is available in
>> libsystemd-daemon-dev, and the current package as a dependency on
>> systemd rather than libsystemd-daemon-dev. It might be right but it
>> does not feel right.
> I would suggest you ask the dovecot maintainer why he build-depends on
> systemd rather than libsystemd-daemon-dev if it is in fact for the
> reasons I listed. I gave a suggestion as to why he would do so, as well
> as a reason for why systemd.pc is not in its own package, but as I am
> not the maintainer of dovecot and there's no way for me to actually
> know, short of asking, which you can just as easily do yourself.