[
https://issues.apache.org/jira/browse/LUCENE-6229?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14324759#comment-14324759
]
Robert Muir commented on LUCENE-6229:
-------------------------------------
I think a lot of the problem i have with the current API is that it is so
undefined, for example when getChildren() or freq() are allowed to be called.
In some cases like ReqOptSumScorer, we can work around the issue partially, but
it makes code more confusing (it calls score() in its freq(), which might
invoke a deferred advance() on the optional part). And it doesn't really fix
the issue, if you navigate to the "leaf" nodes you will find the optional ones
positioned on the wrong docID often. I think the optimization in this case is
simple and effective, so it would be a shame to limit our performance because
of this API.
On the other hand, I like the idea of being able to navigate the scorers tree,
at least for debugging. If the API was completely "transparent" it would still
meet that use case. So the main question is, what are the other important use
cases? Do they justify limiting our performance or completely specializing
potentially many scorers to support, or is there a more restrictive API that
would still work.
> Remove Scorer.getChildren?
> --------------------------
>
> Key: LUCENE-6229
> URL: https://issues.apache.org/jira/browse/LUCENE-6229
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Adrien Grand
> Priority: Minor
>
> This API is used in a single place in our code base:
> ToParentBlockJoinCollector. In addition, the usage is a bit buggy given that
> using this API from a collector only works if setScorer is called with an
> actual Scorer (and not eg. FakeScorer or BooleanScorer like you would get in
> disjunctions) so it needs a custom IndexSearcher that does not use the
> BulkScorer API.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
