:: [DNG] Packaging (was Systemd Shims…
Page principale
Supprimer ce message
Répondre à ce message
Auteur: T.J. Duchene
Date:  
À: dng
Anciens-sujets: Re: [DNG] Systemd Shims
Nouveaux-sujets: Re: [DNG] Packaging
Sujet: [DNG] Packaging (was Systemd Shimss, Init scripts in packages, possibly Mission Creep)
I know I have said this before, but I just wanted to say that both
threads are really the same issue. I think the overarching problem that
both Debian and Devuan have is the very same problem: packaging. From
my little corner of the world, every disagreement so far seems to be a
package problem. The only difference is that Debian is aiming toward
systemd, while Devuan is aiming toward System 5.

1. When on Debian, forced to use systemd or a System 5 + Systemd shim:
package problem because systemd is the package default. Solved with
optional packaging.
2. Potentially missing System 5 Init scripts missing for Devuan
developers. Solved by packaging System 5 compatible scripts.
3. Can't include new versions of gnome or other packages in Devuan.
Solved by systemd optional packaging.

Devuan v1.0 is pretty much going to be a clone of Debian Jesse, and
that's fine, but I think it should stop being a clone there. Devuan
needs to be more than just "Debian without systemd". I think that the
Debian packaging standards need to be reworked (or completely abandoned)
so that Devuan does end up like Debian. If Devuan doesn't do something
different, the upstream work for removing systemd with each release is
likely to just going to keep multiplying to the point of breaking.

Devuan developers are most likely (correct me if I am wrong) rebuilding
the Debian packages "as is" and it is very likely (given the team size)
that will be the norm for the foreseeable future. So they are neither
going to actually test every package, nor do they have the human
resources to turn that around any time soon. Devuan has to depend on
Debian to do the heavy lifting.


What I propose is that in the future Devuan support and test only a
basic subset of the Debian archive as "Devuan." I think that should be
only what is currently the "Debian system base". Anything past that is
simply "sauce for the goose" and supported in a defacto fashion in a
separate repo. I'm sure this might seem like a lack of ambition or
pessimism to some. I think of it as a focused goal with the best
chance for real success. This is not to say that the Devuan Team has
been idle: even the existence of an Alpha 2 is impressive.

All that Devuan really needs to provide is a Debian base system without
systemd. Everything else is secondary to that. At least, it should be.
Everything else can be achieved by adding layers on, but that "Devuan
base" is still needed. Worrying about Gnome, XFCE, or even X11 is
really a secondary goal.