DuckDB LTS and Python Ecosystem

[paultiq]

It's unclear what the 1.4.x LTS plan means for older Python versions and newer Python packages that drop support for them.

Specifically:

• Will 1.4.x support the latest versions of key packages, like Pandas 3.x and PyArrow 22, when they release?
• Will 1.4.x continue to support 3.9 through the end of LTS period, even if its EOL?

Background

• Python 3.9 is EOL this month
• Numpy 2.3.4 dropped 3.9 & 3.10
• Pandas is dropping 3.9 & 3.10 in 3.0: BLD: Drop Python 3.10 support pandas-dev/pandas#60059
• PyArrow (along with adbc) 22.0 will drop 3.9: 22.0.0-rc0
• (edit: add pytorch): PyTorch 2.9.0 just shipped 3.14 support... but: "The minimum version of Python required for PyTorch 2.9.0 is 3.10. We also have 3.14 and 3.14t available as preview with this release."

* Numpy & Pandas adopted https://scientific-python.org/specs/spec-0000/, which is a little more aggressive than the Python lifecycle.

[evertlammerts]

This is a good point. We haven't formulated a specific strategy on any of this yet. Python 3.9 being EOL is a good catalyst, I'll give this some more thought.