[MRUNIT-198] Serialization is missing in MockMultipleOutputs

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.1.0
• Fix Version/s: 1.1.0
• Assignee: Lars Grote
• Reporter: Lars Grote
• Created: 31/Jan/14 13:20
• Updated: 06/May/14 21:56
• Resolved: 04/Mar/14 18:59

Description

Hi,
with issue MRUNIT-13 MockMultipleOutputs was introduced. Which is great! Unfortunately the inner class MockRecordWriter doesn't serialize the Object and therefore isn't storing a copy of the Object but the Object itself. I would suggest to use org.apache.hadoop.mrunit.internal.output.MockOutputCollector instead of the inner class. This Collector does store a copy of the Object and I see no point in having more or less the same Collector/Writer twice.

Another thing that bugs me, is that MockMultipleOutputs requires you to use Comparable Objects in your MR Jobs, and I don't see for what reason this restriction is imposed on me.

I'll provide a patch for this soon and would be glad if someone can comment on it.
Cheers, Lars