I took a look at this and it seems like compiler plans have changed as hive updates and we haven't updated our test outputs with those compiler plans. I think, long term, it makes sense for us to leverage the upstream Hive test infrastructure and use it to run the tests on our fully distributed cluster. That way, we can directly leverage upstream tests and output assertions. Let me take a look and see how much work it is and if it can be done real quick in Bigtop 0.6 time frame. If not, I would be inclined to either do the cosmetic task of updating the test outputs or disabling these tests in Bigtop 0.6 scope and pursue the task of leveraging upstream tests and framework post Bigtop 0.6.