[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

`systemd --system` as a viable way out of the systemd debate?

Dear all,

as probably most others, I am deeply unhappy with the current state of affairs.

All sides have compelling arguments, which means, to me, that it would
be a benefit to all involved if there was a commonly accepted
Maybe there's still room for rough consensus[1], however unlikely.

Has anyone actually tested the viability of running systemd in non-PID-1 mode?
If yes, does this work and would it continue to work?
If yes, is there any hard commitment from upstream in this regard?

Assuming the answers to all of the above is yes, would this be a
viable option for the DDs who want to enforce loose coupling?

>From what I understand from reading, not testing, using this approach
would allow anyone to use systemvinit, OpenRC, or whatever; treating
systemd as a fat dependency, similar to how recent desktops started
depending on databases, etc.

There would be several benefits to this:

* _All_ other init systems would be fully supported, not only a
possibly shifting set of "at least two"; which may or may not invite
another GR down the road
* logind et al could be leveraged by anyone wanting to
* An implicit, or explicit, new expectation would be set: If you bring
a new init system with new features into Debian, it needs to be able
to run as a daemon

If this sounds like a sane middle ground, I can try to hammer out a
new Amendment; hopefully without moving discussion/voting period.

I would prefer a common agreement with everyone voting FD, but that's
simply not realistic as things stand now.


PS: If no one steps forward to actually test `systemd --system`, is
this something we could bounce up to the TC? After all, their opinion
is, hopefully, still valued as in Debian's best interest.

PPS: Thanks in advance for avoiding red herrings, unrelated issues, or
plain yelling in this thread.

[1] https://tools.ietf.org/html/rfc7282

Reply to: