Yes, i think that the debate makes no sense too - for me, Spark is no Hadoop competitor its rather complement.

Spark does not need Hadoop - you can run it also with Mesos or in local mode..