apache / datafusion-python

Apache DataFusion Python Bindings

Home Page:https://datafusion.apache.org/python

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Bump maturin version to satisfy conda-forge constraints?

charlesbluca opened this issue · comments

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
Availability of this package on conda-forge for each Python minor version is dependent on there being an associated maturin package built against that minor version - in the case of python 3.12, there are no builds of maturin that satisfy >=0.15,<0.16, blocking the ability to publish packages for 3.12 (conda-forge/datafusion-feedstock#44 (comment)).

Describe the solution you'd like
Bumping the maturin version constraints to something that contains at least 1.2.3 would give us access to 3.12 builds, allowing conda-forge packages of this project to be published for python 3.12.

Additional context
Discussion around this is surrounding conda-forge/datafusion-feedstock#44.

#725 is merged and upgrades maturin on main.

Does conda-forge pull from main? Or do we need a new release to test if that fix solves the issue?

Does conda-forge pull from main? Or do we need a new release to test if that fix solves the issue?

conda-forge pulls the sdist for a versioned release from PyPI, but I don't actually think the maturin constraint contained in this code forces us to that maturin version; for example, in conda-forge/datafusion-feedstock#44, we are rebuilding using the 34.0.0 source code, which constrains maturin>=0.15,<0.16, we are still able to pull a newer maturin version and build successfully with it.

In short, I think we should be safe to move to the updated constraint in #725 to unblock some of the older conda-forge that have been blocked:

And then continue using this updated constraint for future conda-forge builds moving forward.

As a side note, would you be interested in being added as a maintainer of the linked feedstock? This would give you write access to the repo as well as notifications of these past build attempts.

I don't have a good mental model for conda and its ecosystem, but I just started watching the feedstock repo so maybe I'll learn a little.