Add section on winding order

[felixpalmer]

No description provided.

[alasarr]

Related to #46

[cholmes]

I'm out of my depth here, don't think I've ever dug into winding order. Hoping @tschaub or @hobu sounds in on #46 from the geojson perspective.

[cholmes]

I'm out of my depth here, don't think I've ever dug into winding order. Hoping @tschaub or @hobu sounds in on #46 from the geojson perspective.

[Jesus89]

From this #46 (comment), this covers the planar edges case, but not the spherical edges case. I would make this explicit in the spec. Regarding the spherical case, we can leave it undefined or with an explicit definition too.

[alasarr]

From this #46 (comment), this covers the planar edges case, but not the spherical edges case. I would make this explicit in the spec. Regarding the spherical case, we can leave it undefined or with an explicit definition too.

I think @felixpalmer proposes to force this winding order for both cases

[cholmes]

I think @felixpalmer proposes to force this winding order for both cases

Yes, but I think we also need to specify the orientation rule. It probably makes sense to follow MS SQL, and Google BigQuery and 'interpret the side to the left of the line as the content of the ring'.

[jorisvandenbossche]

I commented on the issue for the more general discussion (#46).

But specific comment on the text in this PR: can we avoid using the term "right hand rule"? I know the GeoJSON spec uses this, but that term is very confusing, as other sources use "left hand" to specify the same .. (eg the cited bigquery and MS SQL sources)

[alasarr]

Thanks @jorisvandenbossche and @cholmes. @felixpalmer I think makes sense to specify that to be more explicit.

[tschaub]

Late to the party, but I agree that this is good to specify. I think mentioning the "right-hand rule" is ok (as some people may know it), but that it also makes sense to add some more detail (something like "the interior of the polygon is to the left of the boundary as you traverse vertices in order").

Inevitably, consumers of the data will do orientation checks before rendering, topology operations, etc. (as the data gets passed through structures that don't keep track of or require a specific orientation), but I still think it improves things to have the format specify how polygon rings are to be interpreted.

[alasarr]

@tschaub this is one of the latest PRs of v02.0. I've added your commit suggestion. Could you add your stamp here?

[felixpalmer]

I agree on the clarification here, it makes it more accessible to wider reader base