starting Fedora Server SIG

[Thorsten Leemhuis]

On 14.11.2008 19:09, Jeremy Katz wrote:
> On Fri, 2008-11-14 at 18:54 +0100, Thorsten Leemhuis wrote:
>>> I'm not saying we keep everything forever, I am saying we might consider 
>>> thinking a bit more before deciding to implement a brand new system 
>>> as solution to an existing problem.
>> That IMHO is a general problem in Fedora and IMHO one of the biggest 
>> problems of Fedora.
> But it's also one of the often-posited strengths of Fedora -- we push
> the edge forward which in the end benefits everyone

Well, I don't think or meant that we should slow down. But we afaics 
should sometimes offer the old known to working stuff as alternative 
when new shiny long term replacement is not a fully working replacement 
yet for 99.5 % of the users.

But yeah, it's hard to draw a line where a "alternative" is wise/needed 
and how much work it's worth to keep it running. I for one would have 
said that in the Juju case a alternative would have been wise and worth 
the work for one release. Otoh is imho was good that we didn't offer a 
alternative for the X-Server 1.5 to make the proprietary drivers work -- 
but I'm sure a lot of people will disagree with the latter.

>> One examples from the last two years: the new firewire stack JuJu. It 
>> afaics is a improvement now, but it was a bit to early when we shipped 
>> it as the "one and only" Firewire stack.
>>
>> Fedora as a whole IMHO should act a bit more like the kernel developers 
>> that have that "no regressions in new kernels" as goal -- they don't 
>> reach that goal completely but they are quite close afaics.
> 
> Kernel developers might argue about the success there.  It definitely
> doesn't match a lot of experiences I see from blogs, bugzilla, random
> other reading of the internets :/

Well, you have a good point, but they are afaics noticeable better then 
Fedora.

Cu
knurd