[ 
https://issues.apache.org/jira/browse/LUCENE-7845?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

David Smiley resolved LUCENE-7845.
----------------------------------
    Resolution: Fixed

In the original optimization I had considered degenerate circles and rectangles 
to be a point.  But they aren't quite identical to a point, so I rolled back 
that aspect.

> spatial RPT optimization when query by point or common date range
> -----------------------------------------------------------------
>
>                 Key: LUCENE-7845
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7845
>             Project: Lucene - Core
>          Issue Type: Improvement
>          Components: modules/spatial-extras
>            Reporter: David Smiley
>            Assignee: David Smiley
>             Fix For: master (7.0)
>
>         Attachments: LUCENE_7845_query_by_point_optimization.patch
>
>
> If the query to an RPT index is a 2D point, or if using 
> NumerBrangePrefixTreeStrategy / DateRangePrefixTree (Solr DateRangeField) if 
> the query is a grid cell (a common date range unit like some particular day), 
> then we can do some optimizations, especially if the data is pointsOnly.  If 
> the data is pointsOnly the strategy can return a TermQuery, if the data isn't 
> then we can at least tweak the prefixGridScanLevel.  This is motivated by two 
> scenarios:
> * indexing polygons and doing lookups by a point (AKA reverse geocoding)
> * indexing date instances and doing date range faceting. Solr's code for this 
> has a fast path for a TermQuery, although more is needed beyond this issue to 
> get there.
> _This development was funded by the Harvard Center for Geographic Analysis as 
> part of the HHypermap project_



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org

Reply via email to