Incompatible with google-cloud-bigquery-storage>=2.0.0

[andrewbrown248]

In the 2.0.0 version of google-cloud-bigquery-storage, they dropped the bigquery_storage_v1beta1 module for just bigquery_storage. This causes the following line to fail importing with google-cloud-bigquery-storage>=2.0.0.

https://github.com/pydata/pandas-gbq/blob/d251db03b159447331ac9ae63e13d295d75bad70/pandas_gbq/gbq.py#L20

Later, when trying to use the storage api, the following error is thrown since the module was unable to be imported.

ImportError: Install the google-cloud-bigquery-storage and fastavro/pyarrow packages to use the BigQuery Storage API.

Dropped bigquery_storage_v1beta1 module:

(env) abrown:~$ pip list | grep google-cloud-bigquery-storage
google-cloud-bigquery-storage  2.0.0    
(env) abrown:~$ python
Python 3.7.9 (default, Aug 27 2020, 21:59:41) 
[GCC 7.3.1 20180712 (Red Hat 7.3.1-9)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from google.cloud import bigquery_storage_v1beta1
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: cannot import name 'bigquery_storage_v1beta1' from 'google.cloud' (unknown location)
>>> from google.cloud import bigquery_storage
>>>

[tswast]

Thank you for the report! The temporary fix is probably to release a patch release with the correct top bound dependencies in setup.py.

Long term, it'll make sense to ensure it works with the 2.0 package, as that's where improvements will be made.

[tswast]

Looks like we don't actually define google-cloud-bigquery-storage as a dependency in setup.py (even though it's needed for best performance). I'll check the pandas-gbq conda feedstock and make sure we define a maximum version there.

[tswast]

It's also not defined in the feedstock (perhaps something to update), but is set to limit to less than 2.0dev in google-cloud-bigquery

https://github.com/conda-forge/google-cloud-bigquery-feedstock/blob/3bf0d97f7e88c01b66421768183b7bf3b8ed7842/recipe/meta.yaml#L119