[postgis-devel] Vote to bump minimum GEOS from 3.3 to 3.5 - PostGIS 2.4.0

[Greg Troxel]

"Regina Obe" <lr at pcorp.us> writes:

> GEOS 3.5 has been out for a while and a lot of our newer features in PostGIS
> 2.3 and 2.4 rely on at least GEOS 3.6.
>
> Looking around, it looks like quite a few package maintainers for 2.3 are
> running GEOS 3.5.0 so I don't think we can bump any higher than that so
> close to release time.

(Not PSC of course, and I've been paying less attention than usual
lately due to Things Other Than PostGIS :-)

FWIW, in pkgsrc we have geos 3.5.  For quite a while it was because (I
think -- am a little fuzzy) osm2pgsql didn't work with 3.6, and now
osm2pgsql doesn't need geos at all.  Now geos is still at 3.5 only
because I haven't gotten around to updating.  Which is a long way of
saying that I feel behind, and am at 3.5 anyway, so it seems like a
correct conclusion that packaging systems that are even sort of up to
date are at 3.5.

Requiring 3.6 I would object to (now), because of the apparent compat
issues from 3.5 to 3.6 and the releative recency given the
non-triviality of the update.  (I realize you explicitly didn't propose
that.)

Also, as a general thing, not really about postgis, I think changing
requirements close to a release should be avoided.  It's natural to be
doing cleanup prompted by realizing a release is coming, but changing
the rules and especially ripping out the old code to cope with the old
dependency versions (since otherwise there would be no gain from
changing the min version?) has some risk, and after that a full round of
testing on all systems/arches is in order.  Given that 2.4 seems to be
alpha, making the change and having another alpha seems fine.  (I
realize I'm on one tail of the norm here.)


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 162 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20170822/6ce8ec25/attachment.sig>