MB-66210: Point and MultiPoint changes #18
Conversation
|
@Likith101 it seems there're a few tests failing within It'd be good if you can address those alongside the github workflow you're about to add here. We can get that merged in first and rebase your sequence of commits over it - so commit validation is run over every pull request. |
|
@Likith101 I've fixed the imports and added github actions for a tests workflow with #23 . Let's address the tests failures that I mentioned above on that commit separately, get that in and then move ahead with your commits. |
|
@Likith101 I've upgraded and addressed the imports of our golang/geo fork with #24. Let's get this in first and rebase your commits over it. |
|
Rebase time :) |
- Switched out all cell operations to point operations
- All point comparisions will use approx equal which has an epsilon of 1e-15
- LineStrings and MultiLineStrings will project point onto themselves
and compare the distance between points for intersection
- Polygons and MultiPolygons use shape indexes with a Closed Vertex Model
which means the boundaries and the vertices are considered part of the shape
- Circle will use distance from center to identify intersecting points
- Envelopes will use bound checks to identify intersects
- New Circles and Envelopes will now init themselves before returning
- Added appropriate test cases
Note:
Projection onto a segment calculation might not be completely accurate.
Projection of point (1,1) onto the line (-2,1) amd (2,1) returns a point that
has a distance of 50m from (1,1). Unable to verify if this distance is actually
accurate or not.
|
I like @Likith101 's recommendation of limiting or refraining from API changes to |
Note:
Projection onto a segment calculation might not be completely accurate. Projection of point (1,1) onto the line (-2,1) and (2,1) returns a point that has a distance of 50m from (1,1). Unable to verify if this distance is actually accurate or not.
Converting the point to a cell and then doing an intersects with the linestring still gives a a false which leads me to believe that the distance it is giving is actually correct.