long term support release

[Dan Williams]

On Wed, 2008-01-23 at 12:10 -0500, David Mansfield wrote:
> On Wed, 2008-01-23 at 12:57 -0300, Horst H. von Brand wrote:
> > David Mansfield <fedora at dm.cobite.com> wrote:
> > > On Tue, 2008-01-22 at 20:24 -0800, Andrew Farris wrote:
> > > > David Mansfield wrote:
> > > > > I'm fairly new to this list so if this is flame-bait, then I apologize.
> > > > > I was wondering whether there is any possibility of having the
> > > > > occasional 'long term support' (LTS) release of Fedora (say one every
> > > > > two years or something) so that users can settle down with the distro
> > > > > and actually become productive with it.  
> > > > > 
> > > > > Say the LTS cycle is one release every two years (every fourth Fedora
> > > > > release), and that the 'long term' for support only lasts for two years
> > > > > (which is pretty short to use the term long for, I realize), then there
> > > > > would only be one LTS release, and also the most current release to
> > > > > worry about at any given time.
> > > > > 
> > > > > If there is simply not enough teampower to do this, then that's
> > > > > understood.
> > 
> > > > That is essentially what was tried with the fedora legacy project
> > > > (supporting eol fedora releases for a longer term) but there was not
> > > > enough interest and support to keep it going.  It did support RH9 and
> > > > FC releases up to FC5 I think?
> > 
> > > Almost the same.  A few differences:
> > > - that project was 'glued on' to an existing process instead of a part
> > > of it
> > 
> > No... it was one of the projects of the (not-RedHat) Fedora almost from the
> > start. They supported pre-Fedora Red Hats.
> > 
> > > - they came into the game with a number of releases to support already
> > 
> > True.
> > 
> > > - they wanted to support every release
> > 
> > The idea was to support only those versions where interest was high enough
> > to support the longer term maintenance. Each version had its fans, but none
> > had enough longer term interest (say, more than 6 months after official
> > EOL) to keep them going. Perhaps the latest Red Hat (9) had a bit more, but
> > I suspect that had more to do with the name change than any objective
> > reason.
> > 
> > > I think Fedora LTS would be:
> > > - planned and built into the Fedora cycle and finally implemented
> > > - only releases planned in advance to be LTS releases would be LTS
> > > - there would only be one (or two) outstanding LTS releases at a time
> > 
> > As was offered, propose a SIG and gather people (lots of them!) to do the
> > (hard, non-glamorous) work.
> 
> If such a proposal was made, it would involve changes to the workflow
> and cycle management of the core releases, even if the bulk of the
> support work is to be done by a separate group.  Would such a proposal
> have a chance?

I wouldn't expect it to have that many changes on the core cycle,
because core would simply keep going on 6-month releases, while the LTS
team could continue to handle the backports for the LTS release for
another year after normal support was discontinued after about 12 - 15
months.

Moving from a 6-month release cycle to a 1 year release cycle for even
just one release isn't going to fly.  Keeping the core release cycle
constant and extending one particular release update lifetime might.

Dan