One of the things I think OpenNLP and Lucene are doing well is really expanding their test planning, coverage, etc. (see https://cwiki.apache.org/confluence/display/OPENNLP/TestPlan1.5.1 and http://wiki.apache.org/lucene-java/TestPlans) I think as we get closer to 1.0, we should start to implement more of this, too. I've started on this at https://cwiki.apache.org/confluence/display/MAHOUT/Testing but would appreciate the help in fleshing it out, assuming others think it is worthwhile.
Thoughts? -Grant
