fthevenet opened a new pull request, #12212: URL: https://github.com/apache/lucene/pull/12212
This PR aims to address issue #12211: Searches made via DrillSideways may miss documents that should match the query. A more detailed explanation of the issue and the reasoning behind the fix can be found in the report linked above, but it basically boils down to the fact that the `score` method in `DrillSidewaysScorer` results in more than one consecutive call to the `matches` method exposed by the `TwoPhaseIterator` instance without re-positioning the iterator first. This in turn, makes the matching of documents erratic, and lead to more or less subtle issues in searches from the end user's view point, where documents that should match there query sometime don't appear in the result list (but may still do if other factors such as the inner type of query resulting from parsing, caching, order in which docs where indexed, etc...) I propose solving the issue by initializing the position of the scorer by call nextDoc on `baseApproximation` instead of `baseIterator`, which should produced the expected result regardless of the type of iterator. In fact, looking back through the history of this code, it feels to me that calling nextDoc on baseIterator is a left over from before two phase iterator where introduced, and should have been changed then. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
