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

Re: systemd-sysv/shim in testing

Sven Bartscher wrote:
> As brought up in this[0] thread, systemd-shim became incompatible with
> the latest versions of systemd and the alternative dependency was
> removed.
> While this is something I can totally understand, I don't think that
> systemd should migrate to testing without the possibility to use
> systemd-shim instead.
> My reasoning about this is as follows:
> Testing should be as close to a state in which it can be released as
> possible (I don't have a citation for that and just remember that I did
> read that somewhere, so please correct me if I'm wrong).

That's an accurate description of testing.  systemd-shim isn't currently
in a releasable state, but that's entirely on systemd-shim to fix.  (The
current round of incompatibility appears fixed in unstable, along with
the inclusion of cgmanager, though there are apparently still a number
of bugs in cgmanager; see
https://lists.debian.org/debian-devel/2014/07/msg00957.html .)

Along the same lines, systemd also currently has a pair of RC bugs on
it, though both have patches that just need uploading.  I feel certain
that'll happen long before it affects the jessie release.

Testing propagation does take installability into account, but
systemd-shim is installable in testing; it just doesn't serve its
intended purpose of supporting systemd-less graphical desktop systems,
hence why it has an RC bug that the current version in unstable should
address once it propagates.

Based on the RC bug policy for releasing testing as stable, either
systemd-shim will have that bug fixed by the time a stable release rolls
around, or if something goes horribly wrong then it'll be removed from
testing, and either way testing will then be releasable.

> Further we are not planning to release jessie without systemd-shim (at
> least I hope so, again please correct me if I'm wrong).

I would assume the systemd-shim maintainers would prefer not to see
jessie released without systemd-shim, hence why they've uploaded a new
version fixing its RC bug (along with the new upload of cgmanager).
However, systemd-shim isn't a distro-wide release goal; it's a single
package with a well-known alternative available.  I don't see any
particular reason why systemd-shim needs any more special treatment than
other packages; either it'll be part of the release or it won't.  At the
moment, it seems quite likely that it'll be part of the release.

We need to decouple the success of systemd and systemd-shim.  Each one
needs to succeed or fail on its own merits.  The systemd maintainers
provided a months-long delay for systemd-shim to catch up to 208 (which
is not the current upstream version), but I certainly hope future
versions of systemd do not incur further delays waiting for alternative
init systems to catch up.

- Josh Triplett

Reply to: