[
https://issues.apache.org/jira/browse/LUCENE-7158?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15220795#comment-15220795
]
Robert Muir commented on LUCENE-7158:
-------------------------------------
+1 but as Nick suggests, it should be forced to be consistent with e.g.
bounding box generation.
We had the problem of multiple haversins before and using different constants
is just as bad. I think if a user does a range query vs sorting vs a lucene
expressions, they should not see inconsistencies just because we don't organize
things properly.
To me this is far more important than any particular value the constant may
have.
> Haversin should use the earth's mean radius, not its max (equitorial)?
> ----------------------------------------------------------------------
>
> Key: LUCENE-7158
> URL: https://issues.apache.org/jira/browse/LUCENE-7158
> Project: Lucene - Core
> Issue Type: Improvement
> Reporter: Michael McCandless
>
> Across our spatial modules we seem to disagree about the earth's radius when
> we model it as a sphere.
> I think in our haversin implementation we use equitorial (maximum) radius,
> but maybe in spatial3d we use the earth's mean radius.
> I think mean makes more sense: the earth is actually a squashed sphere, so
> it's polar radius is shorter than its equitorial radius.
> I think it's important, when we model the earth as a sphere, that we pick one
> radius and try to use that one consistently?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
