[
https://issues.apache.org/jira/browse/LUCENE-8447?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16571600#comment-16571600
]
Adrien Grand commented on LUCENE-8447:
--------------------------------------
Thanks Nick, it looks great. It looks like INTERSECTS now always creates dense
bitsets as opposed to filling a sparse DocIdSetBuilder (the one created in
LatLonShapeQuery is unused). I see how it's helping with code reuse but maybe
we should try to improve it in follow-ups. Regarding your comment about whether
the inverse optimization makes sense with DISJOINT and WITHIN, I think it has
limited benefits indeed. I wonder whether we should drop this optimization
entirely, as well as maybe the other one that checks whether all documents
match, in order to make shapes easier to maintain since I don't think these
optimizations would often kick in in practice. I suspect it will at least help
to remove them temporarily while we are iterating, even if we eventually decide
to add them back once the code stabilizes.
> Add DISJOINT and WITHIN support to LatLonShape queries
> ------------------------------------------------------
>
> Key: LUCENE-8447
> URL: https://issues.apache.org/jira/browse/LUCENE-8447
> Project: Lucene - Core
> Issue Type: New Feature
> Reporter: Nicholas Knize
> Assignee: Nicholas Knize
> Priority: Major
> Attachments: LUCENE-8447.patch
>
>
> This feature will add support to {{LatLonShapeBoundingBoxQuery}} and
> {{LatLonShapePolygonQuery}} for searching all indexed {{LatLonShape}} types
> that are {{WITHIN}}, or {{DISJOINT}} to, the target query shape.
> {{INTERSECTS}} remains unchanged.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
