Re: Bug#727708: tech-ctte: Decide which init system to default to in Debian.
On 29/10/13 23:51, Svante Signell wrote:
> On Tue, 2013-10-29 at 00:51 -0700, Steve Langasek wrote:
>>> (Also, do remember that any decisive outcome other than “support
>>> multiple ones including systemd” and “systemd-only” will need to
>>> lead to the removal of GNOME from Debian.
>> Absolutely not true. As Tollef mentions in his follow-up, one of the
>> options is to fork logind and maintain it.
>> This is not an improbable outcome. Logind is a good interface, and there's
>> a lot of value in continuing to use it, regardless of what init system we
>> choose. If Debian chooses to ship upstart as the default, I will almost
>> certainly be inteested in making sure that logind continues to be
>> well-supported on top of upstart, in both Debian and Ubuntu. The work to do
>> this on Ubuntu has so far been straightforward; while there are some
>> technical hurdles in the future for making logind continue to work on
>> non-systemd systems, these are known issues and not at all insurmountable.
> Some more systemd-login0 dependencies:
> apt-cache rdepends libsystemd-login0
> build-rdeps libsystemd-login-dev
> (not only [linux-any])
> gnome-disk-utility also depends on libudisks2-dev which depends on udev,
> which is linux-any/built from systemd?, so this one might be impossible
> outside linux anyway.
> weston is Architecture: linux-any but
> isn't it supposed to run on other architectures in the future?
That should probably be any and not linux-any, but it makes no difference at the
moment as wayland only builds on linux. When wayland is ported (there were
patches upstream to port it to *BSD but they needed some more work) then I see
no reason for weston not to work on BSD or Hurd.
The systemd integration in weston is optional and could be disabled on !linux
when the time comes, but until wayland is ported there's no point in further
Same for gnome-disk-utility. The systemd bits are optional and could be disabled
on !linux, but there are other bigger issues here.
> Looks like systemd is creping in everywhere.
Because it's useful and easy to integrate with, so people take advantage of
that. I'm not sure what your point is.
> See also the problems with gdm3 not starting #724731, systemd-login0 and
> libpam-systemd. That bug hit me too, on a brand new box as reported
That is a bug, and a grave bug, that I hope to have time to look at soon. I
don't see how that is relevant at all.