I think we should only document the "DAG author friendly" approach (env var
or otherwise) and hold the existing PR until it is ready. My 2c.

Reply via email to