On 11 March 2017 at 14:58, Klaas Freitag <freitag(a)opensuse.org> wrote:
Hi,
I am happy to see that there is some interest in the idea :)
Am 08.03.2017 um 13:05 schrieb Richard Brown:
>
> While I love and respect the work that the Medical and Education teams
> have and continue to do, I think it would be safe to say that their
> experience by running their projects as 'derivatives' ultimately ends
> up being too much work for them - and then backfires on the openSUSE
> Project as a whole as we often saw less direct contributions in the
> area of Education and Medical in the main distributions, which then
> causes a spiral of endless work..
I am not clear what you mean here. What is it that makes a 'derivate'?
I am not fond of seeing many separate openSUSE distributions. We have
Leap and Tumbleweed with a comparable level of quality, and I fear
having more distribution offerings with the name 'openSUSE' gives the
project a negative reputation - at best one that has a confusing
message about what it offers, at worst, a reputation for bad quality
when the 'derivatives' do not meet the same level of quality of the
official 'mainline' openSUSE distributions.
That said, I do understand the appeal of a 'derivative' that is
specifically focused on more narrow use cases - see openSUSE JeOS for
one such example which I think is done right.
In this case, and in the way I'd prefer for all cases, all of the
packages are in the main Leap/Tumbleweed distributions, have all been
quality checked through the usual processes to get them there, and
therefore the only responsibility for JeOS is to polish, package and
rebadge the distribution for that specific usecase.
This is a lot easier to do, especially in the long term, precisely
because a lot of the heavy-lifting of the integration is taken care of
as part of the traditional Leap/Tumbleweed processes.
Other derivatives, that have dozens, or hundreds, of additional,
untested, unintegrated packages in their spin, give me cause for
concern, especially when they do not implement processes and standards
approaching the mainline Leap/Tumbleweed distributions, but still wish
to use the openSUSE name. They also need a consistent and often
increasing commitment to keep them running, as that large overlayed
codebase will need continued rebasing as Leap and Tumbleweed move on.
I see any 'derivative' that distributes a significant number of
packages that are not in the main distribution at significant risk of
failure over the mid to long term.
The current Invis server builds based on openSUSE Leap
and has currently 29
specific packages on top, as one can find in
https://build.opensuse.org/project/show/spins:invis:stable
These are packages are mostly specific to Invis, such as the setup and
special applications which are not in the interest of a generic distribution
(I suspect). Most dependency packages are already in the underlying distros
aka Factory I guess.
I think that this sounds like a proper way of doing it, or not?
invis is a mostly good example of a derivative done right. 29 packages
is certainly not bad. They invis team have proven themselves more than
capable of maintaining that level of deviation for many years now.
But I do not think it's ideal - I would like to see if the SMB team
could look into getting that down to as close as 0. It sounds like
Stefan is on the same page. This is all about integration and
encouraging good proven practices.
Another aspect that should be discussed is that the
Invis server based on
openSUSE currently has it's own branding and it's own setup- and
adminstration tool. I don't think that is a problem, do others?
I don't think it's a problem. But on the flipside, I'd love it if
'mainline openSUSE' was as good as invis for SMB out of the box ;)
Apart from these technical things, I believe that a
huge block of work will
be to do documentation that is neccessary, but even more to do marketing
around this initiative. The nicest project does not help if nobody knows
about it. And the group of pot. users is very hard to reach.
+1 Agreed
Maybe we
should schedule a first meeting to start the Team.
Stefan and me will meet on Chemnitzer Linuxtage tomorrow, maybe somebody
else is around? But indeed the oSC 2017 would be a great opportunity to do
the first 'official' meeting and maybe kickoff?
Sadly I'm not at CLT, but I will be at oSC and most certainly want to
help this initiative in whatever ways I can.
--
To unsubscribe, e-mail: opensuse-project+unsubscribe(a)opensuse.org
To contact the owner, email: opensuse-project+owner(a)opensuse.org