Rolling Release Ubuntu?

[Serge E. Hallyn]

Quoting Daniel Llewellyn (daniel at bowlhat.net):
> 
> 
> > On 6 May 2017, at 12:32, N. W. <nw9165-3201 at yahoo.com> wrote:
> > 
> > So why even do releases in between (16.10/17.04./17.10 for example)?
> > 
> > You're doing a 16.10/17.04./17.10 release (which most users do not use anyway) and then in addition to that also backport lots of things to 16.04. So much wasted/duplicated effort. It would be much better if you would rather do one rolling release and concentrate on that one.
> 
> I’d be interested to see these statistics that back-up the claim of "most users" not using intermediate releases between LTSes. While 51% of users would count as “most” that would also mean several million users (I’m guessing here) using intermediate releases. While it is appropriate to discuss the potential for a rolling release I think that arguments in favour, which are based on assumptions of “most users”, need to be thoroughly researched rather than bandied about liberally without hard facts.
> 
> Daniel Llewellyn

As a user I like the idea of a rolling release.  I like the idea of basically two
tracks for each LTS - a 'stable' and a 'devel'.  We are sort of getting that from
backports, and snaps take it further.

However, the intermediate release serve a very important function - cadence.  It
gives us a nice, predictable timeline for planning incorporation of upstream features,
and gives users a nice timeline as well.  Ubuntu has always been somewhat unique in
this regard, in a refreshing sense.  We shouldn't take losing that lightly.

What has presumably diluted this is the shortened support time for intermediate
releases, which in turn was due to the addition of longer lifetimes for LTS.  Oh,
and of course the physical planning sessions also greatly contributed to the
usefulness of the intermediate releases.

Anyway years ago the advantages of the release cadence were much extolled upon,
and it seems worth considering how to get back those advantages rather than
becoming like everyone else.

-serge