[
https://issues.apache.org/jira/browse/SOLR-12326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16519493#comment-16519493
]
Yonik Seeley commented on SOLR-12326:
-------------------------------------
One part of the solution is for the request merger to know if a shard has more
buckets. If it knows the exact amount of over-request used, then it can figure
it out. This is a little more fragile though, and I could envision future
optimizations that dynamically change the amount of over-request based on
things like heuristics, field statistics on that shard, and results of previous
requests. For that reason, I'm planning on just passing back more:true for
field facets that have more values.
> Unnecessary refinement requests
> -------------------------------
>
> Key: SOLR-12326
> URL: https://issues.apache.org/jira/browse/SOLR-12326
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: Facet Module
> Reporter: Yonik Seeley
> Priority: Major
>
> TestJsonFacets.testStatsDistrib() appears to result in more refinement
> requests than would otherwise be expected. Those tests were developed before
> refinement was implemented and hence do not need refinement to generate
> correct results due to limited numbers of buckets. This should be detectable
> by refinement code in the majority of cases to prevent extra work from being
> done.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
