[ 
https://issues.apache.org/jira/browse/LUCENE-7154?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15219824#comment-15219824
 ] 

Karl Wright commented on LUCENE-7154:
-------------------------------------

Yup, [~rcmuir], thanks, I see it.

Geo3D needs to split non-convex polygons into multiple convex polygons, and I'm 
basically trying to figure out how much work I should put into determining 
whether or not a given provided hole has any intersection with a given convex 
polygon.  That's question #1.  I could use *all* holes for *each* convex 
polygon, of course, and that's what I'll probably do for starters.

The other issue I've got is this business of clockwise vs counterclockwise 
ordering being meaningful.  For Geo3D I just need a point that I know is inside 
or outside.  The internal API asks the user to specify the index of a point 
that is convex (outside the two adjoining points if you connected them).  I 
haven't yet found an algorithm that maps to clockwise/counterclockwise points, 
though.  How are you handling this elsewhere?




> Add support for polygon holes to Geo3D
> --------------------------------------
>
>                 Key: LUCENE-7154
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7154
>             Project: Lucene - Core
>          Issue Type: New Feature
>            Reporter: Karl Wright
>
> Real-world polygons (e.g. from ESRI) have holes in them.  We need polygon 
> support in geo3D that works in the same way.
> The proposal would be to change the GeoConvexPolygon constructor to include a 
> number of GeoPolygon inputs, each of which would specify a hole.  Then, the 
> GeoPolygonFactory.makeGeoPolygon() method would also need to accept a similar 
> list of GeoPolygon hole descriptions.
> This change is likely to be fairly complex because of the already tricky 
> algorithm used to create convex polygons from non-convex ones, implemented in 
> GeoPolygonFactory.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to