Question: does the connection using JDBC must be async?

[burtcorp]

Question: does the connection using JDBC must be async?

[iconara]

The most common use case for this driver is to use the JDBC API to run a query and get the results as one operation. JDBC is a synchronous API and this driver makes interaction with Athena synchronous even though the Athena API is asynchronous.

However, this driver also provides access to the execution ID, which means you can do a mix if synchronous and asynchronous if you want. See "Getting the query execution ID from a ResultSet" in the readme for an example.

[RFayez]

Thank you so much iconara for your quick feedback and help.

I have one more question, In the "Getting the query execution ID from a ResultSet" section, the resultSet will not return until the query is completely finalized, So query execution id in result set metadata will not be used for checking if the query execution is finished or not in such case. is my understanding correct?

[iconara]

Yeah, you're right, I misremembered the implementation. The Statement#execute call does not allow access to the execution ID before the result is available, so the mix of sync and async is not possible. The reason we added ResultSet#getQueryExecutionId was that we wanted access to statistics and metadata, which is only available after completion anyway.

If you need access to the execution ID while the query is running I think it would be an interesting feature to add. I think you could add a AthenaStatement#executeAsync that returns a subclass of Future<ResultSet> that would also have a property that provides access to the execution ID.

I don't have the bandwidth to work on anything like that right now, but I can write a spec if you want to take a stab at a PR.