On Wednesday 07 January 2009 08:55:50 John Wang wrote: > Paul: > > Our very simple/naive testing methodology for OrDocIdSetIterator: > > 5 sub iterators, each subiterators just iterate from 0 to 1,000,000. > > The test iterates the OrDocIdSetIterator until next() is false.
At LUCENE-365 there is an old TestDisjunctionPerf1.java, in which the various scorers iterate by increasing the document nrs by a given number. Iirc the tests there use prime numbers for that, mostly for a somewhat realistic test of the priority queue. It will probably need some care to make it work again, quite a few things have changed since then. The main point to take away from it is to also test the performance of BooleanScorer, which has the best performance I've seen so far on disjunctions. > > Do you want me to run the same test against DisjunctDisi? No need for that I think, the DisjunctionDISI there is still based on basically the same priority queue that Disjunction...Scorer uses. Regards, Paul Elschot
