[postgis-devel] Vote to bump minimum GEOS from 3.3 to 3.5 -PostGIS 2.4.0 - downgrading requirement to 3.4

Regina Obe lr at pcorp.us
Wed Aug 23 10:09:21 PDT 2017


Well as I feared we'd be bitten by our own tools.

It appears some of our bots (I call them Sandrobots - boooo strk)  only have
GEOS 3.4, and I'm too tired to try to figure out how to upgrade them.

So I'm going to downgrade the requirement to GEOS 3.4 and then when we flip
to  PostGIS 2.5 flip to up requirement to 3.5 again and upgrade the docker
images on the Sandrobots.

strk I am very NOT happy.

We have about 2-3 weeks left to go before release, so perhaps it's best not
to be too ambitious skipping 2 GEOS releases.

Thanks,
Regina

-----Original Message-----
From: Greg Troxel [mailto:gdt at lexort.com] 
Sent: Tuesday, August 22, 2017 8:01 PM
To: Regina Obe <lr at pcorp.us>
Cc: 'PostGIS Development Discussion' <postgis-devel at lists.osgeo.org>
Subject: Re: [postgis-devel] Vote to bump minimum GEOS from 3.3 to 3.5
-PostGIS 2.4.0


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

> Now as to upping the GEOS version while we are in alpha.  Normally we 
> try to catch this before release but didn't even notice that we 
> allowed 3.3 as our minimum since we had a glaring warning that you 
> should be using GEOS 3.6 (or was it 3.5, I forget).

I didn't mean you shouldn't take the change.  It sounds sensible.  I just
meant that you should view it as significant, and realize that everyone
needs time to test.  Hence my notion of cycling to alpha.  But if it's
multiple weeks to release, that seems fine.  I am just reacting to previous
history on other projects where I wake up to a big change and a release in
the same 24h period.  But it may also be that it's of near zero consquence
for people testing with up-to-date packaging systems that are all at 3.5 or
higher.

I also don't follow the notion that there are lots of people out there
actually running geos 3.4 or earlier and having performance problems but
that those same people won't have problems if the requirement is changed.
Perhaps the notion is that all of them are behind in some not-ok way and
this will make them deal, and stop bug reports that shouldn't be filed, so
all in all it's good for the community.  If so, that sounds totally
reasonable.





More information about the postgis-devel mailing list