[
https://issues.apache.org/jira/browse/LUCENE-7167?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15223287#comment-15223287
]
Robert Muir commented on LUCENE-7167:
-------------------------------------
That is fine, I just want us to make progress on simplifying the API.
As far as the edge case / quantization handling, again I think we are doing the
complete wrong thing here:
My problem is right here:
https://github.com/apache/lucene-solr/blob/master/lucene/spatial3d/src/java/org/apache/lucene/spatial3d/PointInGeo3DShapeQuery.java#L126-L135
We truncated the users data but we should not try to "correct" for that, it
just makes things complicated for no good reason (and since we throw out edge
cases, all this complexity IMO is basically untested). We decide to use a lossy
encoding, that is fine, but we should not try to "correct for it" at query
time: we just have to accept that but then treat it as if the user handed us
that truncated value from there on out and not allow any more lossiness/error
to sneak in.
> Re-engineer or remove all of geo3D
> ----------------------------------
>
> Key: LUCENE-7167
> URL: https://issues.apache.org/jira/browse/LUCENE-7167
> Project: Lucene - Core
> Issue Type: Task
> Reporter: Karl Wright
>
> Geo3D has led to a lot of consternation because it has a relatively open API.
> The task is to either drastically restrict it or remove the package entirely.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
