Both Hadoop 2.0.0 as well as Windows support seem to be a high percentage of the Jira issues we are now facing. The first big problem I see is that we have no windows testing system. This is annoying as I feel their is no end in sight for windows issues.
We can fix problems like this: https://issues.apache.org/jira/browse/HIVE-3483 But we have no method of coverage testing other then hoping the windows user diagnosed the problem right and hoping they will confirm the fix works. The same looks to be true for hadoop 0.23.0, as the build has never completed successfully. (correct me if I am wrong) https://builds.apache.org/view/G-L/view/Hive/job/Hive-trunk-h0.23.0/35/console Unless someone can convince me otherwise I am under the opinion. That we should stop focusing on the endless stream of Jira requests from these platforms until we have a valid way to test them. In other words 1) put a high priority on getting the 0.23.0 build green 2) stop accepting windows patches till we have some way to test them. Opinions?
