Fedora lifetime and stability

[Bill Davidsen]

Serguei Miridonov wrote:
> On Thursday 08 November 2007, John Summerfield wrote:
>> Mikkel L. Ellertson wrote:
>>> Serguei Miridonov wrote:
>>>> However, before starting a discussion about this I would
>>>> like to ask, if this topic was discussed earlier.
>> To death.
>>
>> I'm sure it was
>>
>>>> but can somebody point me any deep analysis which really
>>>> proves that current one year lifetime and half-year
>>>> release period is the best for Fedora?
>>> Here is a section of a post on Fedora Philosophy:
>>>
>>> The Fedora project does not pretend to be *production
>>> server* centric. It does not even pretend to be
>>> *production server* friendly. The personality of the 
>>> Fedora project is fast paced, (b)leading edge, leaving the
>>> past behind quickly. It is a great proving ground or test
>>> bed for current technologies. It is fun. It will never
>>> have the stability or extended support that a server class
>>> distribution does.
>> Translated, "rolling beta." In return for your access to the
>> latest technology, you can expect cuts and bruises.
> 
> Then don't name "rolling beta" as stable because it is 
> missleading.
> 
>> If you want a longer life, go look at other solutions.
> 
> Look, I'm running Linux since 1994 starting with Slackware then 
> switched to Red Hat and Fedora. I have Linux on both home 
> computer and in my office. I always liked the fact that with 
> every new release the system became more and more stable and 
> usefull. That was before F7. Upgrading from FC5 to F7 wasn't 
> disaster, of course, but this was just because of my knowledge 
> of the system - using Linux more than 12 years makes 
> difference. USB drives did not mount, no problem, we'll do it 
> manually. Kernel did not park heads before switching power of 
> laptop off, well, modern drives use their kinetic energy to 
> remove heads. I could imagine the perception of inexperienced 
> user who for some reason decides to try Fedora as his first 
> (and last?) Linux distribution.
> 
> In July-August most F7 problems were resolved (for my system, 
> at least). The normal life has just started, but now F7 has 
> only 6-7 months to live? This is what makes me just rise my 
> hand and ask.
> 
> My remarks are not to offense developers and maintainers. I 
> myself was a maintainer of a kernel driver and I know what it 
> cost to keep things alive. I started this thread having just 
> one thought in mind - improving Fedora, at least, to return 
> the stability that Red Hat and Fedora had in the past. This is 
> why I suggest to have one release an year, allow more time for 
> testing before the release and extend the lifetime at least 
> for two years.
> 
> If someone wants new and cool bleading edge software, there is 
> always a testing version of Fedora, so long term lifetime 
> isn't a problem. Even some newest packages can be backported 
> to current test updates.
> 
> Actually, I'm not going to continue this discussion. I wanted 
> just to share my thoughts. I know that I'm not alone. For 
> example, here 
> http://www.oreillynet.com/linux/blog/2007/03/where_fedora_went_wrong.html 
> is also said enough, on both sides. And my opinion is that 
> Fedora will only win if testing period and release lifetime 
> will be at least twice longer.

Having just seen such a problem in FC8, I have to agree to some extent. 
Solving the problems by removing and reinstalling packages is my usual 
solution, although I have had to manually run some RPMs, because the rpm 
package ignores the force option and insists it knows better than the 
user. I do see that as a failing of the package, at some point there 
needs to be a way to move forward.

But while I might like to see longer support (as in security fixes, at 
least) for each release, I fail to see how Fedora can "win" by doing so. 
It's hard to see what having more people not paying for the product does 
as a benefit, while the only reason I update most of my systems is 
security, and if I could get security releases for two years, I would be 
testing less new stuff.

There was some mumbling about value to stockholders in this thread, I'm 
a stockholder and I think that FC is a reasonably cost effective testing 
program for new stuff. That justifies its existance, and I think we have 
to be content with that.

Note: I don't see Fedora and Ubuntu competing for the same users, so 
there's not much "win" there, if I didn't run Fedora I'd probably go to 
CentOS for most things, certainly for servers.

-- 
Bill Davidsen <davidsen at tmr.com>
   "We have more to fear from the bungling of the incompetent than from
the machinations of the wicked."  - from Slashdot