Re: Proposal - preserve freedom of choice of init systems
Thibaut Paumard writes ("Re: Proposal - preserve freedom of choice of init systems"):
> I wonder whether this GR has the following corollary:
As the author of the TC text (which Matthew has simply adopted), I
think I can clarify this. What I'm about to say will come as no
surprise to anyone who followed the relevant parts of the TC
discussion.
> "Packages shipping services or daemons must ship whatever is required to
> start and stop said service by any init system introduced in Debian in
> the past and future", which I think would be bad.
No. It only says software may not "require _a_ specific init system"
(emphasis mine). The text does not rule out a package supporting only
some subset of the available init systems - so long as that subset has
more than one member.
In practice so long as a package supports more than one init system,
making it support all init systems is a feasible (or even
straightforward) task.
> Also, is that OK for a package to Recommend a specific init system
> rather that Depend on it?
It doesn't talk about Depends or Recommends. It talks about whether
the program works (or doesn't).
I think that "degraded operation" is exactly what one would expect
with a violated Recommends.
Ian.
Reply to: