I started looking at some of the aggregate code (in connection with https://issues.apache.org/jira/browse/HADOOP-1231) and I noticed that the abacus examples in contrib didn't make it to the main examples when the abacus code was moved. Also, AggregatorTests in org.apache.hadoop.mapred.lib.aggregate isn't actually a unit test - how does it get run? I guess these things need remedying - shall I open a Jira?
Tom
