1. Right now if you use the distributed search servers, the QueryFilter plugins are executed on each server machine. Would it make sense to have the client execute the plugins once, and then dispatch the Lucene query object (instead of Nutch query object) to the server machines?
2. When using distributed search, I believe that docFreq() is handled incorrectly. Each server will perform its own docFreq() calculation instead of taking the true docFreq() of the entire index across all machines. This throws off the scores. I'm having trouble thinking of a clean solution to this problem. Any ideas? Andy
