rmuir commented on PR #12089: URL: https://github.com/apache/lucene/pull/12089#issuecomment-1420793280
> 100%. The issue here is that `TermInSetQuery` gets rewritten to a `BooleanQuery` because there are fewer than 16 terms, so it doesn't have a chance to "self-optimize" to use doc values. We can fix this by not eagerly rewriting to a `BooleanQuery`, but I held off doing that for now. So this is "easily" fixable I think. Well right now, I'm not seeing any justification to mix up concerns and pile all into one query. Currently overall, IndexOrDocValues has the best performance: it is slightly slower in one case but massively faster in that case. Also there are some problems with the benchmark, i tried to wrestle with it but there is some serious noise/gc/something going on. if i just rearrange order of tests numbers change dramatically. I am still willing to optimize the one case where IndexOrDocValues might do better, but I honestly don't feel I need to "play defense" any more. I think we should use IndexOrDocValues. -- 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
