Adventurous updates? (was: Re: Stable Release Updates types proposal)

[Michael Schwendt]

On Sun, 14 Mar 2010 17:42:51 +0100, Kevin wrote:

> What we "adventurous updates" folks 
> really want is non-disruptive non-conservative updates. No need to be 
> conservative as long as you don't break anything.

It's not that updates "don't break anything". The less conservative, the
higher the risk. The larger and the more numerous the changes in updates,
the more likely they break something.
Unless you have the testing resources to ensure a certain level of quality
for the updates *and* all dependencies, which is doubtful. Or else more
bugs would be found in development *and* in updates-testing than in
stable updates, when *many* more users start using the updated software.

Bugs slip through, even embarrassing ones. And apparently, one of the
reasons for these discussions is not only that there is no policy for
updates yet. But that high-impact bugs in some Fedora Updates have slipped
through, because their package maintainers had been willing to take the
risk, and that has prompted some people to try to change that part of
Fedora. How exactly remains to be seen.

> Rawhide is not a solution, as has been explained several times already.
> 
> And many feature updates were, in fact, tested in Rawhide first!

And still you can find enough users who blame Fedora because of its
non-working releases -- and because of the many updates, which don't fix
the bugs they consider the worst.
Most of those users have not participated in the development period and
don't enable updates-testing either, because they fear instabilities when
doing so. They expect others to do that and to release a finished product
that works.
I dunno whether those 70-80% power-users are, who would stop using Fedora
if stable updates tried to bring stability only. Rather I think we're
missing the 70-80% power-users, who *would* use Fedora if it worked better
out-of-the-box and improved over time with updates adding fixes instead of
new risks.
*That*'s why I would like to see me and other packagers retain the freedom
to publish updates as needed, but not so they can produce a
rolling-release in disguise that syncs with development too often.

> When I speak for FESCo, I say so! When I don't say otherwise, I only speak 
> for myself! I'm not a spokesman!

Right, not a spokesman for FESCo, but still a community representative,
elected by parts of the community. Whether you fight solely for yourself
or whether you believe you act on behalf of the community, isn't obvious.

I think more packagers are more concerned about too high hurdles -- when
releasing a bug-fix update (such as testing feedback becoming mandatory
even for niche packages) -- than about a policy on what sort of updates
(and frequency of updates) would be permitted *without* special intervention.

> And FYI, I'm the only one who took your defense during and after the FESCo 
> meeting when those remarks were made about you.

Well, thanks for any defense, but dragging me into the argument in a public
meeting has set off the booby trap.