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

[Stephen Warren]

Arthur Pemberton wrote:
> On Wed, Dec 10, 2008 at 11:57 PM, Stephen Warren
> <s-t-rhbugzilla at wwwdotorg.org> wrote:
>> Arthur Pemberton wrote:
>>> On Wed, Dec 10, 2008 at 11:49 PM, Stephen Warren
>>> <s-t-rhbugzilla at wwwdotorg.org> wrote:
>>>> Kevin Kofler wrote:
>>>>> Jesse Keating wrote:
>>>>>> Seriously, if we could actually just focus on bugfixing for our released
>>>>>> trees, do the new package work in rawhide (and bugfixing of the new
>>>>>> packages there), our released trees might actually stabilize outside of
>>>>>> the heavy handed forced freezes during development.
>>>>> But often it's impossible to fix some bugs without a version upgrade. For
>>>>> example, the upgrade to KDE 4.1 in F9 fixed many bugs in 4.0.
>>>> That's an argument for not having introduced the alpha/beta-quality
>>>> release KDE 4.0 into Fedora in the first place.
>>> And that's an arguement for turning Fedora into something useless.
>> What, usable software is useless? You can't possibly argue that KDE 4.0
>> was a good choice given the need to replace it with 4.1 so quickly.
> 
> 
> First of all, these kind of discussions piss me off a bit, so if I
> fail to be clear on something let me know and I will try to reexplain.
> 
> I never send usable software is useless. I said changing Fedora in
> that way would make Fedora useless. There would be no meaningful
> differentiation between Fedora and other popular distros, and that
> would render it entirely useless to the myself, and I would argue, the
> Linux community. I see very little between what is being suggesting
> and making Fedora "like Ubuntu". The community already has a "like
> Ubuntu".

So, if the idea of Fedora's differentiation is to keep throwing the
latest stuff into the latest non-devel release essentially in order to
always have the latest stuff, or close to it, what is the point of
having releases - why not just have a completely rolling distro?

The only things I can see that a release gives are:

* Creates devel/rawhide as a staging area for the next release, allowing
large upgrades like KDE 4, Perl 5.xx, Python 2.6 to be put in place
piece-py-piece without disrupting users. This could also be achieved by
creating "feature" branches from the distro, making these invasive
changes, then merging back into the single release branch once
everything was done.

* An an installer. Given that Fedora creates beta/preview/RC installers
with minimal locks on devel, it seems that working installers could
still be created even without numbered releases (or perhaps the packages
on the installer lag the top-of-tree by a month or so to allow
stabilization in an installer branch).

Alternatively (to a rolling distro), we could increase the frequency of
Fedora releases. It seems that the main reason people want the latest
stuff thrown into Fedora $latest is because they don't want to wait for
the next release in just 3 months time on average. So, make "the next
release" happen every 3, or 2, or 1 months (-> wait is 1.5, 1, 0.5
months average). I don't know about capacity-wise, but the Fedora
infra-structure certainly seems to have the logic in place now for
managing arbitrary numbers of current releases.