[postgis-users] Proposal to drop support for PostgreSQL 8.3 in PostGIS 2.0

[Paragon Corporation]

> What I did not agree on was your statement that all those
> wanting the latest and greatest PostGIS version would also
> want the latest and greatest PostgreSQL version.  I did
> not have any particular Postgresql version in mind, even
> if the thread was on 8.3.
> Regarding dropping support for 8.3 - I would not call
> backports "default" Postgresql versions.

> Håvard

Håvard,

You bring up a good point.  I guess my thinking is that some distros are
really old.  Like aren't some RedHat versions still tauting PostgreSQL 8.1
as their standard of excellence :)  I can't remember the 
Last time I've used an official distro since they always seem to be eons
behind in the name of stability.

We are not 100% in agreement -- I think Mark Cave-Ayland has reservations,
Kevin has some but fewer, Paul and I have less but we still have some
reservations with dumping PostgreSQL 8.3.

So with that said  and since I am the one who proposed the drop, I would
like to take a quick poll.


We will most likely not be releasing PostGIS 2.0 for at least another 6
months.  Paul or Mark or Kevin can correct me if I misspoke here.
With that said

1) How many people plan to upgrade when PostGIS 2.0 comes out?

2) Of those people that will want to move to PostGIS 2.0, how many are
currently running PostgreSQL 8.3 and still plan to be in 6 months from now?


--- Keep in mind the landscape we are talking about
A) PostGIS 2.0 is going to require a hard upgrade -- no soft upgrade path
will be available.
B) There will be the following versions in addition to 8.3 --> 8.4, 9.0, 9.1
C) 1.5, 1.4, and 1.3 will still be supported on older versions and we will
apply bug fixes etc to those as needed.
D) IF we keep 2.0 -- we might need to drop both 8.3 and 8.4 at the same time
-- e.g PostGIS 2.1 (as the new enhancements coming with index changes etc.
will make it really difficult to hang on to these old versions)

The main reason I want to drop 8.3 is not just because of the lack of
functionality, but just because it is a great effort to regression test
across 4 versions (not to mention the permutations of Oses) given the
numerous minor subtle changes that have happened from version to version.
Its also hard to keep track of what you are allowed and not allowed to do
with each.

So with that said -- supporting 8.3 in PostGIS 2.0 means, to me, slowing our
pace of development for questionable benefit.

Thanks,
Regina


Mike Toews wrote:
> On 22 October 2010 01:16, Havard Tveite <havard.tveite at umb.no> wrote:
>> I do not agree.  I think that many, like me, want the latest
>> and greatest of PostGIS, while running the default Postgresql
>> version for the OpSys.
> 
> Most operating systems that use a package manager have at least 8.4:
> * RedHat/CentOS/Fedora via http://www.pgrpms.org/
> * Ubuntu 8.04 to current, may require enabling backports
> * Debian Lenny via backports
> * Mac OS X via fink, MacPorts
> * Solaris ???
> * Windows requires manual install/updates
> 
> Are there any particular OSs that don't have packages for at least 8.4?
> 
> -Mike
> _______________________________________________