[Mageia-dev] unity on mageia

[Colin Guthrie]

Hi Damian,

'Twas brillig, and Damian Ivanov at 12/09/12 11:33 did gyre and gimble:
> My name is Damian Ivanov. Xiao-Long Chen and me are the maintainers of
> unity desktop environment for openSUSE and Fedora

Thanks for writing and getting in touch!

It's certainly an interesting proposal, but it's really one that opens a
box of many more related questions, some of which I'll try and touch on
below.

1. External Repositories

The "Here is my PPA where I do x, y and z" style approach to software
repositories is always one we've deliberately avoided. It's creates
significant problems for our (relatively) small teams when doing QA and
addressing bug reports. For example if a user is using third party
software and they open a bug report, the Triagers and packagers need to
be aware of that, but it's often something users do not know to
highlight or the decide by themselves that it's "not relevant" (when, in
fact, it often can be due to some obscure library that is loaded
dynamically that messes up the whole stack).

It also creates significant problems on upgrades, if they have upgrade
one package to a non-official version, that then might affect the rest
of the distribution when it comes time to upgrade.

None of the problems here are insurmountable, but they do exist and if
we are to continue to produce a quality product, we need to keep a
strong hold on the reigns with regards to this kind of stuff.


2. If external repositories are to be discouraged then the next obvious
step is to try an incorporate it in Mageia repos directly. However,
unity in particular is a pretty onerous package when it comes to
non-upstream modifications to other packages. This is Canonical's choice
when they designed things this way. I'd rather see more upstream
collaboration, and I'd personally want Mageia to encourage that general
principle, not sign up to the policy that downstream divergence is OK
and something to be supported and encouraged. For this reason, for me
personally, I'd exclude Unity from Mageia official repositories until
the patches can be properly upstreamed. This is a primarily politically
reasoned decision. I don't like divergence and I favour collaboration.


3. Ignoring the above points and thinking more practically, would you
propose to use the same specs for Fedora, OpenSuse and Mageia for the
packages or would you use separate ones? Sorry if this is stated already
and I'm just blind! If they are the same, then I suspect this would
cause even more issues as the distros obviously diverge in their
packaging structure.


4. How do you handle automatic rebuilds when a package changes? e.g.
when we change our official gtk, I presume you would want to
automatically rebuild yours too? Also what if lower level things change,
e.g. automated provides/requires extraction stuff provided at a lower
level. I presume all this stuff "just works" in OBS?


5. Regarding using OBS generally for Mageia, I would not personally be
against it in principle, but there are lots and lots of barriers there.
We do already have a working build infrastructure and we can control it
and have experience of fixing, tweaking it etc. We know it's quirks.
With OBS it would be like starting again. That's not to say it wouldn't
be the "right choice", but it could very easily not be the "right choice
right now". It would take the primary sysadmins to be really
enthusiastic and keen on any transition and know exactly what benefits
it would bring us. Although I can't speak for everyone, I just don't
think there is enough in the way of resources right now for that to be
the case.


6. While I'm sure the web interface is nice, I presume it must link
directly to a RCS in the background? e.g. we'd need it to hook into our
subversion system right now. AFAIUI OBS had a some crazy custom RCS
behind it. Is this now more generic? Can it hook up to subversion and/or
git these days, how would it deal with user authentication?


7. There is a large part of me that wants to avoid adding everything and
anything to Mageia just because we can. Even although this is community
lead, and community driven, if we end up adding everything and anything
we could easily end up being a jack of all trades and a master of none.
We already have, IMO, too many options in some cases. Look at the login
manager world, we have about five or six of those and only one is
actually the one I'd want to use and that's GDM. The reason being that
it is being proactive in genuinely solving the problems of today. We
already have issues with most of the other DMs because they are not
doing PAM correctly or are only half integrated. It's a lot of effort to
keep up with things and unless upstream are being proactive (which many
are not) then it falls to the downstream to pick up the slack. I
personally have no interest in lightdm, slim, lxdm etc. but yet I seem
to have to look at bug reports concerning them. Adding more stuff will
inherently make things harder to deal with unless there is a
corresponding increase in talented and pro-active contributors too!


Just as a final note, I would like to thank you for your interest here.
It's important to note that I'm all for collaboration and for
encouraging contributions to Mageia whatever they are. I don't want some
of my own, personal opinions to put anyone off, but I do want people to
appreciate the broad reasons for such concerns (which I'm sure are
shared by others too) and what would need to be done to address them to
everyones ongoing satisfaction.

All the best

Col



-- 

Colin Guthrie
colin(at)mageia.org
http://colin.guthr.ie/

Day Job:
  Tribalogic Limited http://www.tribalogic.net/
Open Source:
  Mageia Contributor http://www.mageia.org/
  PulseAudio Hacker http://www.pulseaudio.org/
  Trac Hacker http://trac.edgewall.org/