Spark stacks inevitably end up with PySpark though. It's rework for people who already committed to Spark, sure. And for bigger projects that committed to Spark this change isn't justifiable. But for a greenfield project, choosing Spark is just silly today.