[
https://issues.apache.org/jira/browse/SOLR-14800?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17490224#comment-17490224
]
David Smiley commented on SOLR-14800:
-------------------------------------
[~hossman] I'd like your opinion on this, as a long-time Solr committer. There
are a variety of queries that are generally only used in a filter query and
thus their scores don't matter. Nonetheless sometimes they can be in the main
query, and thus their scoring behavior becomes pertinent. I think it's kind of
sporadic which of them scores as zero... though I can say that {{ {!filter} }}
will score as zero, as well as the block-join equivalents. Do you think we
should have them all be 1 now or leave them as they are (some being 0)
irrespective of internal refactoring related to Filter which users don't care
about? Whatever happens to Filter, we can do either way.
> Filter default score should be the provided boost, not zero
> -----------------------------------------------------------
>
> Key: SOLR-14800
> URL: https://issues.apache.org/jira/browse/SOLR-14800
> Project: Solr
> Issue Type: Sub-task
> Components: search
> Reporter: David Smiley
> Priority: Major
>
> I see some inconsistency in what some "constant scoring queries" return for a
> score. Lucene's ConstantScoreQuery yields 1 but it can be boosted (e.g.
> multiplied by whatever). SolrConstantScoreQuery (which wraps a Filter) does
> the same. Filter itself (which was modified to BE a Query years ago) yields
> 0 (not boostable), and so if it's used as a Query (vs only used as a supplier
> of DocIdSet), it's always 0. Nearly all ConstantScoreQuery usages use the
> provided boost (often indirectly via score() which is initialized to the
> boost), and do not hard-code 0. I think we should standardize on this
> approach.
> Note: ToParentBlockJoinQuery (and Child equivalent) use 0 (by code
> inspection) but this seems to be an anomaly.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
