[
https://issues.apache.org/jira/browse/SOLR-16162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17526606#comment-17526606
]
Michael Gibney commented on SOLR-16162:
---------------------------------------
Sure!: currently FilterQuery consults the filterCache on the backend, but
doesn't expose the cached DocSet directly. Rather, it returns a {{Weight}}
derived from DocSet.makeQuery().createWeight(), and the weight returns Scorers
over individual segment LeafReaderContexts. The scorers are then used (via
{{DocSetUtil.createDocSetGeneric()}}) to iterate through matching docs (in the
usual way, as against a "normal" query), collecting into an entirely new
one-off DocSet built via DocSetCollector.
So basically you have a (cached) underlying DocSet, forced through the
Weight/Scorer abstraction to build an entirely new DocSet that's equivalent to
the original underlying DocSet, for every request. The goal of this PR is to
provide a way to bypass the Weight/Scorer abstraction where possible and just
directly use the single cached DocSet instance.
> FilterQuery (`filter([some_query])`) should implement DocSetProducer
> --------------------------------------------------------------------
>
> Key: SOLR-16162
> URL: https://issues.apache.org/jira/browse/SOLR-16162
> Project: Solr
> Issue Type: Improvement
> Security Level: Public(Default Security Level. Issues are Public)
> Components: query parsers
> Affects Versions: main (10.0)
> Reporter: Michael Gibney
> Assignee: Michael Gibney
> Priority: Minor
> Time Spent: 10m
> Remaining Estimate: 0h
>
> There is a substantial practical reason for FilterQuery to implement
> DocSetProducer: the way FilterQuery works, it only consults the cache
> internally for the backing query; the cache is never directly consulted for
> the FilterQuery _per se_. As a consequence, when FilterQuery is used in a
> context that wants a DocSet, currently (without implementing DocSetProducer),
> a full clone of the cached DocSet must always be built via the Weight
> returned by FilterQuery.createWeight(). Implementing DocSetProducer allows to
> avoid superfluous DocSet creation.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org
For additional commands, e-mail: issues-h...@solr.apache.org
