ENH: Support Python 3.11

[EwoutH]

Python 3.11 is expected to be released as stable around October 24th, with many new features including:

• PEP 657 -- Include Fine-Grained Error Locations in Tracebacks
• PEP 654 -- Exception Groups and except*
• PEP 673 -- Self Type
• PEP 646-- Variadic Generics
• PEP 680-- tomllib: Support for Parsing TOML in the Standard Library
• PEP 675-- Arbitrary Literal String Type
• PEP 655-- Marking individual TypedDict items as required or potentially-missing
• bpo-46752-- Introduce task groups to asyncio
• The Faster Cpython Project is already yielding some exciting results: this version of CPython 3.11 is ~ 19% faster on the geometric mean of the PyPerformance benchmarks, compared to 3.10.0.

The state of the Python 3.11 ecosystem is already great:

• ✅ The regular scientific computing stack is in excellent shape. NumPy and Pandas support Python 3.11 and have wheels available, as do SciPy, Matplotlib and Seaborn. Tools like, Black and pre-commit and pytest also support Python 3.11.
• The geospatial stack is build on top of those, and now also gaining Python 3.11 support very quickly
  • 🚀 fiona the CI passes for 3.11, wheel building is being worked on. See tracking issue: Support Python 3.11 Toblerity/Fiona#1087
  • ✅ shapely just released version 1.8.5, with 3.11 wheels on PyPI!
  • ✅ pyproj supports Python 3.11 and has wheels on PyPI
  • ✅ SQLAlchemy idem dito
  • ✅ rtree just released their 1.0.1 release which include Python 3.11 support and wheels!
  • ✅ geopy has a version-independent Python wheel available

GeoPandas is the glue that connects the whole geospatial Python stack. I think it would very well be possible to have full Python 3.11 support in GeoPandas before the official release! (currently planned for October 24th). Would that be a target to aim for?

[martinfleis]

Given geopandas is pure python, it should be straightforward as soon as all dependencies have 3.11 support. If you'd like to figure out a CI env that may test it right now, go for it!

[EwoutH]

Thanks for getting back! I think aside from the GeoPandas dependencies, the CI relying on Conda will be the hold up. Conda is quite slow with adopting new Python versions. I don't think Conda is needed if all dependencies have wheels on PyPI available, right?

I think moving the dependencies and the optional docs/dev dependencies to the pyproject.toml would also help. The number of needed dependencies for each kind of job can be reduced quite a bit probably.

[martinfleis]

I think aside from the GeoPandas dependencies, the CI relying on Conda will be the hold up. Conda is quite slow with adopting new Python versions. I don't think Conda is needed if all dependencies have wheels on PyPI available, right?

Yes, it can be a bit cumbersome initially. You don't need conda to install geopandas so we should also be able to create a CI env without it if we really want. It just adds some complexity to our CI setup I am not sure I am happy about.

I think moving the dependencies and the optional docs/dev dependencies to the pyproject.toml would also help.

See #2512

[EwoutH]

shapely just released version 1.8.5, with 3.11 wheels on PyPI! Only Fiona is left of the direct dependencies!

Conda is working on it, see tracking issue conda/conda#11170.