I am looking for recommendations as to the best Python GIS library currently available based on the following (subjective) criteria:

• Ease of use / Pythonic interface / Documentation
• Power in terms of available features today, and potentially in the future (is it still being developed or dead).

So far I have looked at, in some detail:

• GEOS/OGR/GDAL: So far I have been looking at this library, using this webpage as a tutor. My initial reaction is that while some things have been easy, already I have been running into a few potholes. For example, if I want to create a MultiPoint geometry and determine which points are inside a Polygon using the Union() function, I run into problems, but the Python bindings are just generated by SWIG and all of the documentation is computer generated. I made this post because it struck me that there might be better libraries out there.

• GeoDjango: Well this is again based on the GEOS C/C++ library, but it appears the interface might be intuitive, and possibly better documented. I also have the sense it is being actively updated. I'm not interested in web applications, but this doesn't seem to be an issue?

• Shapely: Again based on the GEOS library, seems to have a Pythonic interface. Not sure about feature support or development status.

Others? Thoughts? Thanks in advance!

In 2011, this is how you check a project's vital signs: https://github.com/sgillies/shapely/commits/master/.

In my honest opinion (disclaimer: I am the lead developer and manual author), Shapely's documentation is right up there with any other GIS software; maybe not quite as good as the PostGIS manual (crême de la crême), but better than ArcPy's, and much better than GEOS's.