[
https://issues.apache.org/jira/browse/SOLR-14164?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17361036#comment-17361036
]
David Smiley commented on SOLR-14164:
-------------------------------------
FYI my current status here is paused, pending some eventual benchmark. A
[conversation on the Lucene dev
list|https://lists.apache.org/thread.html/r58c417c187fdf4fac1aa5b7a79454a813a68cc14299f96b6baa59c9c%40%3Cdev.lucene.apache.org%3E]
pointed to Collectors being faster than TwoPhaseIterator based filtering. If
this is true, then I'm not sure yet what I'll do. I may explore have Solr
detect that a non-cached filter query has a TPI, and if so then wrap it in a
Collector / PostFilter. Maybe even Lucene could do that. That would be a new
issue and it would make merging the current issue "safe" (no perf regression).
> Replace Solr's FunctionRangeQuery with Lucene's
> -----------------------------------------------
>
> Key: SOLR-14164
> URL: https://issues.apache.org/jira/browse/SOLR-14164
> Project: Solr
> Issue Type: Sub-task
> Reporter: David Smiley
> Assignee: David Smiley
> Priority: Major
> Time Spent: 50m
> Remaining Estimate: 0h
>
> Solr has a FunctionRangeQuery. It is constant scoring and also a post-filter.
> Lucene has a FunctionRangeQuery that I added years ago LUCENE-6919. It is
> not constant scoring as it scores based on the wrapped ValueSource, although
> this is easily made the same as Solr's by wrapping in a ConstantScoreQuery.
> And it uses Lucene's TwoPhaseIterator indirectly which is equivalent
> functionality to Solr's PostFilter. Consequently we can remove Solr's
> implementation. Additionally, ValueSourceRangeFilter is a piece of the
> puzzle that can be removed as Solr's FunctionQuery was a wrapper around this
> specifically. FYI Once upon a time, Query and Filter were separate, then
> Filter was made to subclass Query and then Filter was relegated to Solr as
> tech debt.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
