[postgis-devel] PSC vote Request: Make Geos 3.5 the default in PostGIS 2.2

[Paragon Corporation]

> Regina,

> I fear I don't quite understand your original proposal to start with.
> How can PostGIS make any GEOS version "a default"? You can pose minimum
requirements and/or emit warnings during `configure`, yes.
We had this same issue a while back with GEOS 3.2/3.3-- I think it was 1.5
or 2.0 where some portions of upcoming PostGIS relied on GEOS (not yet
released).

GEOS and PostGIS are very closely tied since Sandro is the head of GEOS
development and does a good chunk of the work in PostGIS as well and exposes
functionality in PostGIS he recently releases in GEOS.  So the milestone
Development of the two projects is probably much more intertwined than most
outsiders realize.

Anyway by default, I mean - PostGIS 2.2, WILL allow you to compile with a
lower GEOS, but you need to pass in an extra agreement in configure to do so
just so you know by doing so you are missing some important functionality.
I guess we can just go with a warning, but I personally tend to ignore those
kind of things, perhaps other packagers don't.

> However, to me it seems pretty evident that raising the minimum
requirement to GEOS >= 3.5 is way too premature at this point in time.
Yes I agree making it a minimum requirement is premature, which is not what
I was proposing.  I proposed you can still compile with GEOS 3.4, you just
get a strong urging not to.

>> "Paragon Corporation" <lr at pcorp.us> writes:
>>> This is just for PostGIS 2.2 and the plan is since some very popular 
>>> functions in PostGIS 2.2 will not work without GEOS 3.5, then GEOS 
>>> 3.5 WILL BE released before or very near PostGIS 2.2.

> ISTM the only way you (postgis) can guarantee that is to hold back the
release of PostGIS 2.2 until GEOS 3.5 comes out the gate.
Yes agree here, PostGIS has t to come after GEOS 3.5, not before.


>> Just as PostGIS 2.2 has to be released before or very near release of 
>> PostgreSQL 9.5

> It's PostGIS which depends on PostgreSQL not the other way around. So
shouldn't this rather read: PostGIS 2.2 has to be released *after* the
release of Postgres 9.5? Otherwise, you cannot guarantee compatibility.
PostGIS 2.2 will support PostgreSQL 9.1, 9.2,9.3,9.4,9.5.  and we've been
following 9.5 very closely to make sure PostGIS 2.2 works with PostgreSQL
9.5.
That said even if per chance we release a tiny bit before 9.5 (say two
weeks), and PostgreSQL 9.5 happens to push out a change that breaks our
code, we'd simply push out a micro version 
2.2.1 to fix the issue.  So we don't really need to wait for PostgreSQL 9.5
to be released.  The point of having it ready at the same time or before
9.5, is so people who want to go to PostgreSQL 9.5 aren't held back since
many distros (at least in past) only allowed for one minor version of
PostGIS.
It would be in my mind a huge disappointment if the default that went with
9.5 was 2.1 (since there is great functionality in 2.2, only available for
PostgreSQL 9.5 users) and no special functionality in 2.1 for PostgreSQL 9.5
users.

> I think a "may not depend on unreleased versions" rule avoids the 
> downstream problems when expectations are not met, which is why I 
> think it's important to follow that rule.

> Absolutely crucial, yes.
> Regards

> Markus Wanner

Yes I agree too.  Keep in mind I'm thinking 3 months forward not the
present.  I rarely spend time thinking about the present.
PostGIS 2.2 is at least 3 months from release, so is PostgreSQL 9.5, and
GEOS in theory is much closer, I think can go out within next month
personally. 

Hope that clarifies things.

Thanks,
Regina