[
https://issues.apache.org/jira/browse/LUCENE-7168?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15223327#comment-15223327
]
Karl Wright commented on LUCENE-7168:
-------------------------------------
Hi Robert,
The reason for the 8 implementations is because there are degeneracy
conditions. Specifically, when a dimension is immeasurably small, we still
need to determine set membership but only when it is on the plane that
represents that dimension. it's dicey to use two planes that are right on top
of each other for determining "in set" in that case. There is a degeneracy
condition for a single point as well.
This is independent of the BKD implementation's quantization effects; Geo3D
would want to do this anyway. There are similar issues with bounding boxes
that have degeneracies of various kinds.
> Remove geo3d test leniency
> --------------------------
>
> Key: LUCENE-7168
> URL: https://issues.apache.org/jira/browse/LUCENE-7168
> Project: Lucene - Core
> Issue Type: Bug
> Reporter: Michael McCandless
>
> Today the test hides possible failures by leniently handling quantization
> issues.
> We should fix it to do what geo2d tests now do: pre-quantized indexed points,
> but don't quantize query shapes.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
