With the new approach to our query library API, the underlying runner for distributed data processing is completely separated from how the end user of the library, uses the generated Pandas DataFrames. Our current focus is the Spark runner but we will need a BigQuery based implementation too, i.e., implementing this class.
The new query API is a work in progress and we should focus on BigQuery runner only after we are fairly confident the query API is good enough for the immediate use-cases we have.
With the new approach to our query library API, the underlying runner for distributed data processing is completely separated from how the end user of the library, uses the generated Pandas DataFrames. Our current focus is the Spark runner but we will need a BigQuery based implementation too, i.e., implementing this class.
The new query API is a work in progress and we should focus on BigQuery runner only after we are fairly confident the query API is good enough for the immediate use-cases we have.