[RFC] Releases planning

[Martin Pool]

One specific thing in this that John raised in
<https://code.launchpad.net/~vila/bzr/rm-tweaks/+merge/38240> (rm
meaning 'release manager' not 'remove'):

I think, for us, release candidates are not worthwhile, and they're
just a waste of work and a source of latency.

We do find bugs all the time, but we don't often find critical bugs
that require us to slip the release, and if we do then we have the
window between code freeze and announcement to make a change.  We do
get some good testing of betas, and of code and installers before the
announcement, and I am not convinced there is a large slice of users
who would test rcs but won't test betas.

I think basically we should optimize for the common success case which
is that the .0 release is good.  If we find bugs, we can fix them in
the .1 release, and if we find very serious bugs we can do a .1
release quickly.  Users who are very risk averse can and probably will
just wait a while before upgrading.

-- 
Martin