Mailinglist Archive: opensuse-project (244 mails)

< Previous Next >
Re: [opensuse-project] Don't search developers. Develop them. Events, events everywhere
Hash: SHA1

On 05/09/2015 01:24 PM, Richard Brown wrote:
On 9 May 2015 at 15:14, Martin Schlander
<martin.schlander@xxxxxxxxx> wrote:
Torsdag den 7. maj 2015 10:23:58 skrev Richard Brown:
But technical contributions to the Regular Release are in
decline. Building 13.2 fell on the shoulders of a few people.
We had something like 14 bugs found during the 13.2 Beta

You can't make too many conclusions on the basis of the 13.2
release process.

It was the first release ever based on the new Tumbleweed. It was
always clear it was going to be a bit bumpy. You had SLE12 coming
out around the same time, the testing phase was excessively
short, the roadmap wasn't known until the last moment, iirc a
"13.2" product wasn't created in bugzilla, so a lot of
factory/tumbleweed bugreports will have been actual 13.2 bug
reports. The whole situation was chaotic.

If you dismiss the model of Tumbleweed based openSUSE releases on
this background, you never gave the model a fair chance.

This is a fallacy. While we did rename Factory to Tumbleweed, we
didn't revolutionise the development model for the Regular Release

In fact, we didn't change a thing

Things went into Factory, Factory had a snapshot, the release
process kicked into gear, we pumped out a release

The model for 13.2 was no different than the development model for
every release I've ever been involved in, so I think I can make
conclusions based on the 13.2 release process

and those conclusions really boil down to a few things

1. The quality of Factory/Tumbleweed snapshots are awesome 2. The
vast majority of our contributors interest is in Tumbleweed, not
the Regular Release. 3. Despite every effort to remedy the
situation, the amount of people working on openSUSE Regular
Releases has continued it's decline which we already saw over the
last few releases (just look at vs for a
really stark illustration of how bad it is)

We need a new Regular Release which either a) appeals to a new
breed of contributors who will invigorate the Regular Releases or
b) reduces the amount of work required so we can sustain producing
regular releases alongside Tumbleweed as a rolling release -
because frankly, right now, the status quo is not sustainable

Luckily, the availabilty of the SLE Sources, in my opinion, gives
us an opportunity to kill two birds with one stone. By taking the
Regular Release in a direction that favours stability, we can use
the SLE Sources to make the work required to do that way easier. We
get to shape it based on what we, the community, are able and
willing to do, as brought up by the debate so far. Some bits will
move faster, some bits wont, the bits that move faster will be on
the communities shoulders to maintain, the bits that wont, we get
to benefit from SUSE's work in this area and not compromise on the
quality of what we actually provide our users.

A specific stable distribution for a specific type of audience that
is significantly different from our Tumbleweed user base ->
Invigorated contributors working on a targeted project -> awesome,
a) is handled A whole pile of SLE sources and maintenance updates
meaning openSUSE won't have to worry about parts of the stack
which, frankly, right now we really dont have many people
maintaining heavily (eg. Kernel, GCC, etc) -> Reduced work for the
openSUSE community -> awesome b) is handled

Let me give you an example which illustrates my why I think our
current model just doesn't work, from both a contributors and
users standpoint. For this example, I'm going to use a topical
example of KDE which has been also discussed heavily lately.

openSUSE 13.2 has at least another 20 months of life in it, based
on our Support Lifecycle, assuming we do the next openSUSE release
in November 2015 and another in November 2016 It has KDE 4 KDE 4
will be effectively end-of-life in the eyes of Upstream before
November 2015. That means no more maintenance, no security
updates, nothing from upstream. We're on our own. In the minds of
our contributors (the awesome openSUSE KDE team), this is a very
uncomfortable position. In Tumbleweed, they see it as impossible to
maintain KDE 4 any more, and so they are putting in Plasma 5. And I
think they're right, for Tumbleweed. But look at how noisy our
mailinglists have been because of it.. Quoting Raymond from the
other thread "Why otherwise to upgrade from 13.2, if you get
nothing newer in return." - This mindset "we must ship the latest
of everything, because that's what's awesome" is a very prevalent
mindset in our community. It could practically be the Tumbleweed
development motto. It's why a lot of us get up every day and do
stuff for openSUSE, there's this wonderful ecosystem of open source
stuff and we want to package it up in a distro so we and other
people can enjoy it. And there are certainly a good pile of users
who feel the same way. And so, I think I've just described the
driving force behind Tumbleweed But again look at how noisy our
mailinglists have been because of this mindset We have other users
who crave stability. Who pick openSUSE because we build stuff that
works, not because we build stuff that changes all the time (and
also works). These are the 'long tail' of people we know who are
still using openSUSE 12.x, or openSUSE 13.1 and looking forward to
an Evergreen release in the future. We've seen this come up time,
after time, after time. Every major flamewar the openSUSE community
has gone through, often ends up coming back to this topic, one
group wanting stuff faster, one group wanting stuff slower. We had
it with the release schedule discussions years ago, version
numbering, systemd, moving from KDE 3->4, GNOME 2->3, and now KDE
4->5 And every time we've ended up with compromise solutions.
Compromises are great, but ultimately they're agreements which
piss off both sides equally. Using the desktop transitions I
mentioned above, this often meant the actual contributors, actually
working on the new stuff, reduced their speed of change, held
themselves back, in order to try and satisfy the concerns of others
who felt things was changing too fast. And, frankly, I dont think
that's cool. It's demotivating for the contributors, it doesn't
help upstream projects, in some cases it actually means more work
for everyone involved. It's not ideal. I like the idea of an
openSUSE Project that can keep up, and in fact lead, the open
source world by having a distribution that works, but uses
everything new as soon as it's possible. This is one of the things
that really excites me about Tumbleweed, we're there already and
we're able to take it to the next level now.

So, if we're not compromising on speed because we have Tumbleweed,
why should we compromise on Stability for our Regular Release? Why
not have a Regular Release that, in November 2015, has KDE 4 still
as an option? Users will be happy.

Sorry, but are you not conveniently ignoring the argument you made above

KDE 4 will be effectively end-of-life in the eyes of Upstream before
November 2015. That means no more maintenance, no security updates,
nothing from upstream. We're on our own.
In the minds of our contributors (the awesome openSUSE KDE team), this
is a very uncomfortable position. In Tumbleweed, they see it as
impossible to maintain KDE 4 any more, and so they are putting in
Plasma 5.

SLE does not have KDE, thus an openSUSE "regular release" that is
based on SLE and has a longer lifespan would potentially perpetuate
this problem, if we want to have KDE as an option. Therefore, even in
a "based on SLE" world the KDE team would probably prefer to move to
KDE 5. Thus, I fail to see the difference the difference for the users.

Maintaining it shouldn't be hard if upstream aren't changing
anything. The promise of 'you always get the newest of everything'
might be broken, but this is a promise that is now fulfilled by
Tumbleweed. Maybe we also are able to ship KDE 5 *as well*,
because, with a more stable base, and an expectation from users
that the openSUSE release will move less dramatically, less often,
teams like our KDE team should be able to make sensible choices
about what versions of awesome-upstream-stuff they want to keep in
the Regular Release for a while

The only 'flaw' in this plan, that I totally accept exists, is
that for anyone who feels the current openSUSE regular release is
'perfect', then things might be changing for you in a negative way
However, I'd ask those people, deep down, to identify the key
reason they like the current openSUSE regular release.

- - It works
- - The speed of changes matches what I am willing to deal with for my
desktop system
- - I get new stuff at a reasonable pace such that I can plug in newer
consumer grade stuff and it works

If you're primarily motivated because "every release I get all the
new stuff", then please, use Tumbleweed, and if it's not perfect
for you, help us make it better.

Lets put kGraft into Tumbleweed and my issue with Tumbleweed on my
desktop will go away.

If you're motivated to use openSUSE because "I want a Linux
distribution that just works", then please, help us with this new
Regular Release, in order to make it perfect for you use cases.

This would require kernel developers that are willing to backport
consumer grade drivers onto the SLE kernel.


- --
Public Cloud Architect LINUX
IRC: robjo
Version: GnuPG v2

To unsubscribe, e-mail: opensuse-project+unsubscribe@xxxxxxxxxxxx
To contact the owner, email: opensuse-project+owner@xxxxxxxxxxxx

< Previous Next >