Explicit orientation field for polygon winding #74
Conversation
|
Following discussion in #46 |
|
Remember to update:
|
Co-authored-by: Even Rouault <even.rouault@spatialys.com>
format-specs/geoparquet.md
Outdated
| The winding order of polygons follows the [GeoJSON spec](https://datatracker.ietf.org/doc/html/rfc7946#section-3.1.6). Polygon rings MUST follow the right-hand rule for orientation (counterclockwise external rings, clockwise internal rings). Traversing vertices of rings in order, the interior of the polygon is on the left. | ||
| This attribute indicates the winding order of polygon. Available values are: | ||
|
|
||
| - counterclockwise: the winding order of polygons follows the [GeoJSON spec](https://datatracker.ietf.org/doc/html/rfc7946#section-3.1.6). Polygon rings MUST follow the right-hand rule for orientation (counterclockwise external rings, clockwise internal rings). Traversing vertices of rings in order, the interior of the polygon is on the left. |
There was a problem hiding this comment.
Could be nice to explain a bit more what winding order is. If there's a good web page on it then perhaps just a link to it.
Also I feel like we should start with describing the rule, and then mention that GeoJSON does the same, but not open with 'the winding order follows geojson'.
There was a problem hiding this comment.
then mention that GeoJSON does the same
concur
There was a problem hiding this comment.
Thanks for pulling this together @felixpalmer ! Greatly appreciate your effort to move quickly on this.
I tried to add some clarifying text below.
I think the big decision point at the moment is whether orientation is conditionally required if edges="spherical". At the moment, I'm loathe to rush that in as a hard requirement in the spec; I think most requirements should start as optional and graduate after time and vetting in the broader ecosystem. I think in this case we can leave it optional, encourage writers to set it, and leave it up to readers to demand that it be present (required in practice rather than in specification). That gives more time for others to weigh in that are more familiar with spherical coordinates before making it a hard requirement.
format-specs/geoparquet.md
Outdated
| The winding order of polygons follows the [GeoJSON spec](https://datatracker.ietf.org/doc/html/rfc7946#section-3.1.6). Polygon rings MUST follow the right-hand rule for orientation (counterclockwise external rings, clockwise internal rings). Traversing vertices of rings in order, the interior of the polygon is on the left. | ||
| This attribute indicates the winding order of polygon. Available values are: | ||
|
|
||
| - counterclockwise: the winding order of polygons follows the [GeoJSON spec](https://datatracker.ietf.org/doc/html/rfc7946#section-3.1.6). Polygon rings MUST follow the right-hand rule for orientation (counterclockwise external rings, clockwise internal rings). Traversing vertices of rings in order, the interior of the polygon is on the left. |
There was a problem hiding this comment.
then mention that GeoJSON does the same
concur
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
Co-authored-by: Joris Van den Bossche <jorisvandenbossche@gmail.com>
Co-authored-by: Joris Van den Bossche <jorisvandenbossche@gmail.com>
|
I think we shouldn't mix 0.3 release with adding the orientation field as optional |
There was a problem hiding this comment.
This is looking good to me overall, thanks for the continued effort here!
I think we should be careful about making any recommendation for spherical coordinates when orientation is absent, given the discussion in #46; I think we should just point out is a a good idea to have it and problematic if absent.
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
Co-authored-by: Brendan Ward <bcward@astutespruce.com>
| @@ -32,6 +32,7 @@ | |||
| "geometry_type": ["Polygon", "MultiPolygon"], | |||
| "crs": df.crs.to_wkt(pyproj.enums.WktVersion.WKT2_2019_SIMPLIFIED), | |||
| "edges": "planar", | |||
| "orientation": "counterclockwise", | |||
There was a problem hiding this comment.
This is wrong, as the example data is not actually using CCW rings.
The current example file is converted from a Shapefile, which has by definition an opposite winding order.
(ideally we would have multiple examples so we have both cases covered, but until then we should ensure this is correct for the current file)
|
I opened #80 with a short-term fix for the example file and some small clean-up (I think there was a left-over sentence from accepting an inline review comment) |
No description provided.