Bug#959920: systemctl,elogind: No more co-installable

[Axel Beckert]

Hi Mark,

Mark Hindley wrote:
> I haven't used systemctl myself, but it clearly has similar usage case to
> elogind so it would be ideal if they were coinstallable.

As far as I understand it from the package descriptions (and really
not more than that), they're providing quite some different parts of
the plethora of systemd functionality:

* elogind mostly provides the session manager functionality (i.e.
  systemd-logind), i.e. is only related to providing additional
  functionality around user logins, cron sessions, etc.
* systemctl mostly provides the binary "systemctl" which is related to
  controlling services, i.e. a part of the init system functionality.

For me these are two completely different things which are both united in
systemd, but IMHO don't need to.

> > If the systemd package would provide Provides for at least those
> > interface also offered by alternatives (e.g. systemd-systemctl,
> > systemd-logind), then elogind would just need to have to Conflict with
> > "systemd-logind" and systemd-systemctl would just need to Provide
> > "systemd-systemctl" or similar. And they would be co-installable
> > again, because they both provide replacements for different interfaces
> > of systemd.
> 
> This already partially exists in that the logind and default-logind virtual
> packages are in use to resolve the libpam-systemd libpam-elogind issue.

Ah, right. But this is only for these two packages, not for whatever
exactly separates libpam-elogind and elogind itself.

> However, that does not help in this case. elogind has to conflict
> with systemd as they include some duplicate files.

*nod*

> I have to admit to being less convinced by the systemctl Provides:
> systemd.

Actually, yes, this solution is far from perfect (and probably far
from good, too), but it's still the right way to go IMHO.

As Dmitry wrote: »on what percentage of interface compatibility
warrants "Provides"«. And I think there is no hard limit, but there
are some percentages which IMHO say clearly yes (e.g. 99%) or clearly
"no" (e.g. 10%). It's though very difficult to determine the
percentage values. E.g. is systemctl providing rather 5% or rather 30% of
systemd. ;-)

Actually, In contrary to Dmitry, I would like to discuss this. But not
here. So I'll stop here. The above should only serve as example why I
think that what Dmitry did is the right direction, but not yet really
good. :-)

Oh, and thanks for the pointer to #959174 in the other mail. It hurts
to read the hostility in there. :-(

		Regards, Axel
-- 
 ,''`.  |  Axel Beckert <abe at debian.org>, https://people.debian.org/~abe/
: :' :  |  Debian Developer, ftp.ch.debian.org Admin
`. `'   |  4096R: 2517 B724 C5F6 CA99 5329  6E61 2FF9 CD59 6126 16B5
  `-    |  1024D: F067 EA27 26B9 C3FC 1486  202E C09E 1D89 9593 0EDE