[postgis-devel] Can we put back GEOS 3.5 support in 3.0?

[Raúl Marín Rodríguez]

Hi,

> I had taken it out with the
> assumption we'd skip PostgreSQL 12 cycle and do a 2 year run.

I've always thought that plan was to match 3.0 with 12, if possible.

I have mixed feelings about this. GEOS 3.5 is still receiving commits (last
one in Oct 20, 2018) but, on the other hand, its last release (3.5.1) was in
Oct 2016 which will be 3 years old by the time we plan to release 3.0.
Considering that I expect Postgis 3.0 to be supported for at least ~4 years,
that will mean that we'll need to support an 8 year-old GEOS with its bugs
and differences from 3.7.

I'm personally happy to kill GEOS 3.5 support but I can accept that it will be
a pain in the ass for anyone using a LTS distribution. Nevertheless, those
distributions are already stuck with old GEOS and Postgis. For example Xenial
(Ubuntu 16.04) has GEOS 3.5 and Postgis 2.2, and Bionic (18.04) has GEOS 3.6
and Postgis 2.4.
This means that if you want Postgis 3 and you are using a LTS distro, you
already need to either use some unofficial repo or package postgis yourself.
I don't think that asking to also package a newer GEOS is too much to ask.

I share with Komяpa the feeling that we should push people to also update GEOS
as part of their Postgis update, but personally I might more comfortable telling
people that I'm not going to workaround a bug that isn't present in newer
(major) release, alas that isn't trully supporting 3.5.

So basically I don't have a strong opinion either way, but I'm more inclined
towards dropping it.

-- 
Raúl Marín Rodríguez
carto.com