[postgis-devel] Much slower processing on GEOS 3.9.0 versus 3.8.0 for geodesic area calculation

[Paul Ramsey]

> On Apr 27, 2021, at 2:17 AM, Marco Boeringa <marco at boeringa.demon.nl> wrote:
> 
> - Issue started after upgrading to GEOS 3.9.0 from 3.8.0, but I need to verify if any (dependent) packages were updated as well in the process per Raúl's question. Job to be done by me for which I will report back, although I think it likely that GEOS maintainers should be able to answer this question at least partly themselves: were any dependencies changed to new versions to support the GEOS 3.9.0 release?

GEOS has no dependencies. So no.

PostGIS, on the other hand has lots of dependencies, so if you bumped PostGIS, it's entirely plausable that you'd get a new proj.

> This was a response to a message by me where I received an error message attempting to use a new PostGIS 3.1.1 specific function ("ST_ReducePrecision"), that indicated this lack of GEOS 3.9.0 compilation. Is that a potential source of more problems than just not being able to access a new function definition? Are there any other potential knock-on effects of this, or can we exclude this as being of any relevance to this new performance regression issue?

I find it hard to credit that this could have an effect, but it's sufficiently odd that I cannot discount it entirely. A way to avoid the question would be to just compile from source rather than work with packages.

P