My package's Python version depends on one of its dependencies'. Since it integrates with Databricks, the corresponding runtime environment (DBR) sets a dependency package databricks-connect which then fixes the Python version. That is to say:

DBR 14.3 -> Python 3.11
DBR 15.4 -> Python 3.11

So I'm developing the package using Poetry, but they way I've found it manages dependencies is the other way around... you first fix the Python version, and then adjust the dependencies to match that.

I've tried setting them up with groups and extras, but haven't really solved anything. Most of the time I get conflicts with the dependencies, because the corresponding databricks-connect version conflict with each other.

Is there an approach that works with Poetry, or do I need to look further into other packages such as Tox?

Thank you for your help.