[postgis-users] [postgis-devel] VOTE: Drop PostgreSQL 9.0 support at PostGIS 2.2

Paragon Corporation lr at pcorp.us
Mon Jun 10 02:21:21 PDT 2013


 
Magnus,

> I haven't fully kept track of which other packages do support multiple
versions. But at least both yum.postgresql.org and apt.postgresql.org
support multiple *postgresql* versions. Meaning that if they *do* support
only one postgis version, there might be a problem if you drop support for
older versions of pg - since older versions of postgis don't necessarily
work with the newer versions of pg.

I still fail to see why this is even an issue to bring up to package
maintainers.  
We will still be supporting older versions we started supporting on 1.5,
2.0, 2.1 branches.
Our general policy is that for each minor we release, whatever we stated we
would support we continue to support until postgresql calls EOL on that
minor.
So there should never be a situation where this comes up.

It's more of an issue with when we cut off support for beta versions of
PostgreSQL at current stable PostGIS releases (e..g we did not support 9.3
on 2.0) because there was too much backporting involved. 

1.5 on 9.2 was a greyzone because people tried to build  on it, and pramsey
ported back my fixes in 2.0 on 1.5 before I could smack him.
That was a particularly grey since there were some backward compatibility
issues since 1.5 and 2.0 were big milestone changes.

For betas its hard to know if we'll support or not since we look at the
changes and how frequently it breaks our builds before we decide and also we
end up having to if we are not ready to release by the time PostgreSQL makes
its release.



>> That said this is for 2.2 anyway (and PostGIS 2.1 is still in beta) 
>> and PostGIS group will still be supporting 2.1 and 2.0 on 9.0 .
>> Given 2.2 hasn't even been branched yet, I think this is pretty much a 
>> non-issue since no package maintainers will be compiling against
>> 2.2 yest.  That's why I wanted to get this out of the way so we don't 
>> run into starting supporting anything and then cut it off midstream 
>> while people have already started building against it and thinking 
>> they can use it with their version of PostgreSQL.

> Actually, it would probably be a good thing in general if we could get the
packagers to build against the development snapshots as well - to find bugs
quicker, and get them taken care of before it hits release.
> I'm not sure how feasible it is, but it would be good in general :)

Agreed.  I know devrim (yum.postgresql.org) tries to which is why we called
postgis 2.1 freeze and released a beta versoin.  Keep in mind we as PostGIS
developers haven't started our 2.2 run yet so there is actually nothing to
test there.

> And yes - it's definitely a good thing to get it done early. If it's going
to be desupported, the earlier people find out, the better!
Definitely :)

Thanks,
Regina




More information about the postgis-users mailing list