Answer: Adding a watermark is necessary for a stream-stream outer join.

Let's analyze each option: - **A stream-stream join is not supported in Spark.** This is incorrect because Spark does support stream-stream joins. - **A stream-stream join can only be performed in Scala.** This is incorrect as stream-stream joins can be performed in various languages, including Python, not just Scala. - **Adding a watermark is necessary for a stream-stream outer join.** This is correct because, for an outer stream-stream join, a watermark is essential to determine how long the query should wait for data to arrive, ensuring accurate results. - **A stream-stream join is supported in Spark only if one of the streams has Kafka as its source.** This is incorrect because stream-stream joins in Spark can utilize many streaming sources, not limited to Kafka. - **A stream-stream join cannot be an outer join.** This is incorrect as, unlike stream-static joins, stream-stream joins can indeed be of the outer type. For more detailed information, refer to the documentation on watermarking in stream-stream joins in Databricks and Spark.

Author: LeetQuiz Editorial Team