PyArrow version requirement about databricks-sql-python

[susodapop]

@Tom-Newton pyarrow 10.0.1 is now supported as our v2.2.1 release.

@YS-L I'll look into reducing our pyarrow specificaiton. I think we landed on version 9 because it was the latest at the time we wrote the connector.

[susodapop]

I want to make this work for you but I also can't reproduce your installation problem. For your info: note that our specific requirements were dictated by Poetry so we haven't deliberately excluded anyone and thus far our tests run automatically on Python 3.7-3.11 without issues.

You're welcome to open a pull request that modifies pyproject.toml and poetry.lock to meet your needs. As long as it passes our build checks I'm happy to merge it. But this isn't something we're prepared to do on our own because nothing is broken for us.

[Tom-Newton]

I created a PR #74. It needs a review and for someone to approve running the build checks.

[Tom-Newton]

I have a similar issue but personally I'm interested in using pyarrow 10.0.0. It would be great if this was allowed.

[Tom-Newton]

Thanks @susodapop 🙂 . With 2.2.1 my pyarrow problem is solved but now numpy is an issue. We are on numpy 1.22.4 with python 3.8. Is there are reason that the numpy requirement needs to be so strict?

We are currently using databricks-sql-connector==2.0.4 with numpy==1.22.4 and we haven't noticed any issues.

[Tom-Newton]

I created a PR #74. It needs a review and for someone to approve running the build checks.

@susodapop please could you approve running the github workflows

[susodapop]

Done!