.python-version
always matches the minimum supported Python version.
It’s main purpose is to catch accidental use of too modern syntax
in scripts and functional tests.
We (currently) don’t specify a minimum patch version, so it’s not necessary to do so here. The minor verion is enough.
This also avoids requiring users to keep a potentially unsafe old patch version installed.
The linter CI job used python_build
(part of PyEnv) which requires specifying an exact Python version. The first commit changes it to use the higher level pyenv install
which will pick the most recent patch version it knows about.
The original stated reason for using the lower level python_build
#26716 (comment) was that it allowed specifying CC=clang
which needs fewer resources, but pyenv install
can do that as well.