Fedora Life Cycle

[Paul Stauffer]

On Tue, Oct 23, 2007 at 10:34:08AM -0200, Rodrigo Padula de Oliveira wrote:
> Let's go!
> - - Fedora is released
> - - 1 or 2 months packing and releasing new and necessaries packages
> versions ( now we have 11 months).
> - - 1 or 2 months planning, studying the impacts, creating the migration
> plan and applying it(now we have 9 months)
> - - ohhhh in 9 months we have to do this again
> - - so, we will use CENTOS or DEBIAN.
> 
> That is the problem! Year by year migrating all fedora systems to the
> new version!

Hi Rodrigo,

I understand the frustration you feel in this tension between the benefits
of a fast-moving rate of innovation and the need many organizations have to
not be constantly upgrading all of their machines.  You are not alone in
feeling torn between two choices, neither of which seem optimal for you.  I
am sure there are a lot of other organizations that are in the same
position.

I also note with admiration that you are clearly a person with a lot of
energy and a sincere passion for the Fedora Project.

Therefore I have a few questions for you.  How large an organization is
SERPRO?  What is its annual budget, roughly?  And how many other large
organizations are you aware of that have similar concerns?

The reason I ask is that it seems to me that if an organization with
thousands of employees really depends on Fedora for its daily operations,
and if that organization would greatly benefit from an increased support
cycle for Fedora, then it might easily be worth it for that organization to
hire a small handful of developers specifically tasked with continuing the
maintenance of the distro for a period of time after the Fedora Project has
ceased to support it.  And if this might be a worthwhile investment for
SERPRO, perhaps it would also be worthwhile for a few other large
organizations in similar situations.  Given a handful of organizations, each
sponsoring a handful of developers, it seems to me you could easily
accomplish what you desire, and effectively revive something like the Fedora
Legacy project.  Skip every odd (or even) numbered release, and concentrate
on providing only the most critical updates for your supported release for
one additional year, giving you a total supported lifespan of ~2 years, and
you'd never be supporting more than two releases at once.

Of course, it would take a person with a lot of energy and a sincere passion
for Fedora to organize such an effort, and to convince a few organizations
like SERPRO that this would be a cost effective way to solve their problem. 
Perhaps you know such a person? :)

cheers,
- Paul

-- 
Paul Stauffer <paulds at bu.edu>
Manager of Research Computing
Computer Science Department
Boston University