[
https://issues.apache.org/jira/browse/LUCENE-8039?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16241923#comment-16241923
]
Karl Wright edited comment on LUCENE-8039 at 11/7/17 12:32 PM:
---------------------------------------------------------------
The proposal would be to add another method to GeoDistance:
computeDeltaDistance, with associated implementations and default
implementations.
The delta could be computed one of two ways:
(1) As a perpendicular excursion from the route
(2) As a straight-line addition of another waypoint on the route
The first approach is consistent with everything else we've done here, and is
consistent with the notion of the specified point being reachable primarily
through the route itself. The second approach is probably unrealistic except
when the corridor is rather wide.
was (Author: [email protected]):
The proposal would be to add another method to GeoDistance:
computeDeltaDistance, with associated implementations and default
implementations.
> Geo3d: Need a "distance delta" distance metric for paths and circles
> --------------------------------------------------------------------
>
> Key: LUCENE-8039
> URL: https://issues.apache.org/jira/browse/LUCENE-8039
> Project: Lucene - Core
> Issue Type: Improvement
> Components: modules/spatial3d
> Affects Versions: 7.1
> Reporter: Karl Wright
> Assignee: Karl Wright
>
> The current "distance" method for a path returns a distance computed along
> the path and then perpendicular to the path. But, at least in the case of
> paths, it is often preferable to compute a "delta" distance, which would be
> the minimum straight-line distance assuming a diversion to reach the provided
> point.
> A similar "distance delta" for a circle would be defined as returning a
> number exactly the same as is currently returned, with the understanding that
> the point given would be the destination and not a new waypoint. Similarly,
> the distance beyond the end of a path to the provided point would be counted
> only once, while the distance before the beginning of the path would be
> counted twice (one leg to the point, and the other leg back from that point
> to the start point (or nearest path point, if closer).
> This obviously must be implemented in a backwards-compatible fashion.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
