:: [Dng] Thoughts on Devuan
Page principale
Supprimer ce message
Répondre à ce message
Auteur: T.J. Duchene
Date:  
À: dng
Sujet: [Dng] Thoughts on Devuan
I'm just going to express an opinion here, and I apologize if anyone takes
offense, but I feel it is important that it be said if Devuan is going to
gain ground and the respect of its peers. I think that the best effort
should be made to maintain a friendly dialogue with Debian, especially
since no matter what direct Devuan decides to go, we are still going to be
getting upstream code from them. A lot of the posts I see are
anti-systemd and while this disdain is perhaps will earned, Devuan can't
just focus on a group hate of systemd. I'm not trying to annoy with that
suggestion, only that that is what Devuan might appear to be to someone who
does not understand what we are about.

Devuan should provide a solid statement to the rest of the community about
its intentions if it is
going to be taken seriously on any level. Like Debian, Devuan has to
establish a set of project goals and a standard of quality in order to
attract contributors. Our concern over systemd lacking that standard of
quality is really why we are here. Anyone can take code and fork it, what
really makes a difference is not that we are willing, it is what we do with
the opportunity presented here.

If all we are going to do is repackage Debian to remove systemd and remove
the dependency chains, then so be it, and we should say so. It's simple and
easy to manage. Personally, I think we can do better. Systemd is not a bad
idea. We just don't want to be tied down to it. I
believe that Gentoo Linux has a sane engineering approach: have it for the
people who want it, but it is an optional item.

Debian has done a lot of good things, but has also made a lot of mistakes.
I think that Devuan can avoid most of them if Devuan is bit less stringent
in accepting contributors into the community. The process of joining
Debian can be very difficult and slow, which makes
it harder to grow. Non programmers are especially hard to recruit to work
in documentation and other areas, for example. We have all seen that a
Technical Committee vote can cause a lot of issues and disagreements with
segments of the developers.

I think Devuan could avoid most of this, simply by inviting anyone who
wants to help, but instead of adopting Debian's structure, look at having a
small "core" version. Everything that goes into the "core" is on the
proviso that everything submitted is subject to a knowledgeable group for
QA, but must be kept as small as possible.

The rest of the distribution is really open to what a particular person
really needs, and anyone and everyone should be invited to pitch in. Sure,
you might end up with some packages that are less
than perfect, the the community can sort that out on its own with bug
reports or perhaps an overall grading system. One thing that Debian does
not always do well, and I think that we should devise from the onset is the
ability to replace packages cleanly with different versions. If our concern
is to maintain freedom of choice, the ability to replace any binary package
the end user has issues with is essential.

Lastly, and I think that this is very important, Debian's source build
mechanism is very limited to whatever is in the current tree: say Wheezy.
You can't easily build and integrate code from Testing or any other hosted
archive into your own version. It would be nice to go the extra effort if
it is realistic goal. I'm just suggesting that it should be considered for
review, as an insurance policy against being locked into official archives
- with all the problems that that might entail should some official archive
not be to everyone's liking.


Thanks for reading and sorry about the length. I hope it made some sense.
I'm a programmer, not a politician.

T.J.