What Fedora makes sucking for me - or why I am NOT Fedora

[Kevin Kofler]

Stephen Warren wrote:

> Kevin Kofler wrote:
>> 
>> That's not really feasible, it would be a lot more work to maintain and
>> require a lot more bandwidth and disk space for mirrors. And it wouldn't
>> bring any benefits over the current system.
> 
> The benefit would be waiting less time for significant package updates
> (i.e. new versions for the sake of new features, or significant new
> versions for the sake of bugfixes) if Fedora had a policy that each
> release should be somewhat stable during its lifetime, and updates in a
> release should be minimal/stable/...

But the _current_ system does _not_ have such a policy, so you'd be solving
a problem you introduced yourself.

>> As for a rolling release model, that may be more feasible, but again what
>> are the benefits over the current model? And it'd have the major drawback
>> that users could no longer decide on their own when to upgrade e.g. to
>> KDE 4 (for which right now they have a window of ~7 months).
> 
> I didn't propose that so much because it had benefits, but more because
> if maintainers are free to (and even encouraged, because Fedora is
> bleeding edge, and that's what it takes to be bleeding edge) to keep F
> $latest updated with the latest releases of some/most/all software, then
> we essentially already have a rolling release (it just gets renamed F8,
> F9, F10 every 6 months), so why not just be explicit/honest about it.

We don't already have a rolling release. The policy is clear: if it breaks
things, keep it to Rawhide, otherwise, push to testing and once it's
confirmed not to cause trouble push to stable. (If you think that's not
clear enough, it should be made so.) But these decisions are at the
maintainer's discretion, because those are the ones who best know what
breaks things and what doesn't. (In particular, I don't think we should be
definining "breaks things", it's a matter of common sense.) Forcing an
inflexible policy would cause more problems than it would solve. There are
exceptions to every rule.

        Kevin Kofler