Yep, we could definitely build this into Yellowfin out-of-the-box.. however the documentation says that it only requires CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX if catalog and schema are both null in the call to getColumns()/getTables().

In nearly all of our calls to these functions, we pass the schema defined at the datasource. Does the client have a schema defined at on the datasource?

Or is it just extra slow when both catalog and schema are not specified?, and not as slow when schema is specified?

 

In most of our code we don't pass the catalog to getColumns()/getTables() as catalog means different things for different database systems. Since in SnowFlake, the catalog is the database name, we could potentially put that into calls to getColumns()/getTables().

Is there any speed difference between using CLIENT_METADATA_REQUEST_USE_CONNECTION_CTX and specifying both catalog and schema?