[
https://issues.apache.org/jira/browse/LUCENE-7093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15194210#comment-15194210
]
Michael McCandless commented on LUCENE-7093:
--------------------------------------------
bq. If we there many values (lets say a couple of hundred values) for a single
field, wouldn't the current visit method be better
It's hard to say ... it invokes {{compare}} more than the normal BKD tree
would, and the compare can be more costly than filtering a single value?
Also, I think the visit method only works in the 1D case ... for e.g. the 2D
case, I think it may be buggy because the byte[] values were sorted only by the
first dimension?
I feel like it's best to get a simple, correct, implementation in at first, and
then worry about optimizing for the "many points in a single document" case
later?
> MemoryIndex does not support points
> -----------------------------------
>
> Key: LUCENE-7093
> URL: https://issues.apache.org/jira/browse/LUCENE-7093
> Project: Lucene - Core
> Issue Type: Bug
> Reporter: Robert Muir
> Assignee: Martijn van Groningen
> Attachments: LUCENE-7093.patch, LUCENE-7093.patch
>
>
> I realized this glancing at LUCENE-7091.
> I think this should have points support or else people cannot move off of the
> deprecated LegacyXXX encodings?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
