Would people be open to enhancing the BigtableIO sink to avoid checking if
the table being written to exists in BigtableIO.Write.validate()?

There are a couple reasons I would like to disable the check:
- I would like to create the destination table as part of the flow, with
the schema defined by data read from the source during the flow.
- The machine that launches the flow may not even have the access needed
(both network and IAM permission) to access Bigtable.

Any thoughts?

Reply via email to