BigQuery Storage API integrations should accept either v1beta1 or v1 client

[tswast]

I notice that the v1beta1 API is still used.

python-bigquery/google/cloud/bigquery/_pandas_helpers.py

Lines 577 to 587 in 952164b

	def _download_table_bqstorage_stream(
	download_state, bqstorage_client, session, stream, worker_queue, page_to_item
	):
	position = bigquery_storage_v1beta1.types.StreamPosition(stream=stream)
	rowstream = bqstorage_client.read_rows(position).rows(session)
	for page in rowstream.pages:
	if download_state.done:
	return
	item = page_to_item(page)
	worker_queue.put(item)

This is good, as it keeps packages such as pandas-gbq from breaking, but shouldn't we start thinking about migrating folks to the v1 endpoint? To do that without causing too much disruption, the to_dataframe, magics, and DB-API integrations should accept either v1beta1 or v1 clients for an extended period of time.

[plamut]

@tswast Now that the BQ Storage client is stable, #55 will start using it as a default, and also migrate to v1 stable from v1beta1.

If this would indeed represent to big a disruption, the PR needs to be adjusted to support both Storage client versions. Possibly with a warning if beta client is used?

Also feel free to bring this point up on the linked PR, if you have time, as this feature request directly affects it, thanks!

[tswast]

Yes, that would be a very big disruption. Definitely a breaking change, as it would break pandas-gbq among others.

[plamut]

Noted. We will add this then, and branch on the BQ Storage client type (and if v1beta1 is passed, the old logic for constructing the session will be used).