gautamworah96 commented on PR #12465: URL: https://github.com/apache/lucene/pull/12465#issuecomment-1679412568
Just to add some more context here, we were in fact doing something wrong with our IndexedDISI iterators. At Amazon Search, we have some custom caching logic that shares IndexedDISI iterators for custom fields that all rely on the same underlying fields. What ended up happening was that, in most cases, all fields would always advance to the same doc, but in certain rare cases, we ended up in situations where one field (that shared the iterator) was trying to start from scratch, ie, 0 when other fields had already advanced the iterators to a forward position. Up until Lucene 9.7, we never hit an error because the code was lenient and always returned false if the iterator was trying to access a position that was backwards. However, with PR https://github.com/apache/lucene/pull/12324 this logic was changed and that broke some of our tests. We have now found the bug and have fixed it! Yay! -- 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
