[Pkg-postgresql-public] postgresql 8.2 packaging

[Paul E Condon]

On 2009-03-01_12:31:13, Markus Wanner wrote:
> Hello Gerfried,
> 
> Gerfried Fuchs wrote:
> >  This is what I wanted to raise, too. For a cleaner upgrade path I would
> > suggest that for whatever version we want to have as default in squeeze
> > release we also should have 8.3 still around at that time so that
> > someone can do an upgrade to squeeze, install the new (propably 8.5?)
> > packages and do the pg_upgradecluster.
> 
> Interesting. For how long do you think 8.3 should stay in squeeze? Are
> you voting for providing minor releases (bugfixes) during that period?
> How long do you want 8.3 to stay in squeeze?
> 
> If I understand correctly, you would also vote for having 8.1 around in
> lenny for upgrades from etch, right?
> 
> I don't quite understand the intention, though. If you only want to have
> it around for upgrades, you don't need 8.3 in squeeze at all. You just
> migrate from the currently installed 8.3 from lenny to 8.5 from squeeze
> and then use 8.5 bugfix releases, no?
> 
> >  Additionally, to have people and applications do the upgrade anyway and
> > not stick with the old version having the old version drop the -dev
> > package should do the trick (I'm not sure but I think that was done
> > during lenny development cycle already?).
> 
> Yeah, the -dev packages don't carry the major release version in their
> package name.
> 
> Also note that 8.1.16 is not any older than 8.3.6. And it's highly
> probable that 8.1 it's still in productive use by more people than 8.3.
> 
> >> The Postgres project doesn't currently have a solution to the upgrade
> >> problem (and instead offers pretty good long term support). So people
> >> often use these well maintained major versions. Debian shouldn't prevent
> >> users from doing it that way, IMO.
> > 
> >  Please don't start the same discussions over and over again.
> 
> C?dric asked, and I'm explaining the underlying problem. I'm sorry if
> that's annoying you, but as long as Debian (or Postgres itself) doesn't
> have a solution to this problem, I need to bring it up from time to
> time, so you don't forget ;-)
> 
> > Debian
> > doesn't prevent users from doing it that way at all, as you are the best
> > example for that.
> 
> As stated by Martin and yourself, Debian wants to encourage people to
> upgrade between major releases. It does so by not maintaining older
> major releases of Postgres, thus not shipping security and other fixes,
> which is crucial for productive use. That's what leads me to say that
> Debian "prevents" users from doing it that way.
> 
> But let's stick to your above point, your vote for 8.3 being shipped
> with squeeze. That's an interesting new turn to this discussion, at
> least for me.
> 
> Regards
> 
> Markus Wanner

Markus,
I'm an outsider on this issue of Debian and versions of PostgreSQL. 
I think you are on the losing side of an argument, and deserve to
lose, because your argument is mistaken. Debian packages PostgreSQL
for the convenience of members of the Debian community, not as a
service to the (larger?) PostgreSQL community. 

Recently, there was a thread (I can't find it now) about whether or
not debian-user is moderated or not. Some people, not officially in
management of Debian, offer portals that give access to debian-user.
At least one of these offerings is moderated by the offerrer, The
email writer questioned whether this was 'against Debian policy'.
Well Debian policy applies to Debian operations, not what others do
under other sponsorship. Debian, as a matter of policy, does not
promise to make everyone obey Debian rules. To do so would be a 
truly stupid policy. But Debian does, as a matter of policy, presume
to state, openly, what its policy _is_. As a member of the Debian
community, you can argue your case, but you cannot make Debian do
something that it clearly cannot do, namely enforce Debian rules on
PostgreSQL. Nor are you able to win your case in Debian discussions,
and get Debian to change its support one version only rule. 

To me, there is not a pressing reason for Debian to change on this
issue. There is a big, big world out there. There is plenty of
opportunity for others to step in, and do an Ubuntu for PostreSQL. Do
you want to lead it?  Go find people who are willing to follow. I 
think there is not a Debian policy against proselytizing for volunteers
to another project, so long as the proselytizing does not violate the
rule against commercial advertising, and spam.

To me, the current policy works. It provides a clear boundary to what
a developer commits to when he/she adopts a package. Without this
policy the developer could look forward to a constant struggle trying
to placate both progressives and conservatives. I would not want that.

HTH
-- 
Paul E Condon           
pecondon at mesanetworks.net