[iOS] visibleCoordinateBounds inaccurate when longitude discontinuity around +/-180 is visible on screen

[pietrasm]

Hello,

It seems like visibleCoordinateBounds returns longitude range of -180 degrees to 180 degrees whenever the longitude discontinuity around +/-180 is visible on screen.

Do you have any plans on handling this corner case?

SDK version: 3.2.0rc-1

Thanks,
Michal

[1ec5]

This code should be removed, because a constrain mode of HeightOnly no longer allows world wrapping (#4655).

[1ec5]

There is some question as to what we should return in that case. We could return unwrapped coordinates, in which case you’d see the bounds extend beyond ±180° longitude, but that may be a breaking change: for compatibility with MapKit, this API has always used wrapped longitudes.

[pietrasm]

MapKit returns a single point and size which avoids the issue by design and I think that's the cleanest way of doing it. I think the least breaking change to the current API would be to return wrapped coordinates. For example, if the area begins from 10 degres to the West from the discontinuity and ends at 10 degrees to the East you can return visibleCoordinateBounds.sw.longitude == +170.0 and visibleCoordinateBounds.ne.longitude == -170.0. It still might be a breaking change as there might be expectation that visibleCoordinateBounds.sw.longitude < visibleCoordinateBounds.ne.longitude but it also would be a good indication that the area contains +/-180 longitude. Another idea would be to return two rectangles - one to the left of discontinuity and one to the right.

[1ec5]

For example, if the area begins from 10 degres to the West from the discontinuity and ends at 10 degrees to the East you can return visibleCoordinateBounds.sw.longitude == +170.0 and visibleCoordinateBounds.ne.longitude == -170.0. It still might be a breaking change as there might be expectation that visibleCoordinateBounds.sw.longitude < visibleCoordinateBounds.ne.longitude but it also would be a good indication that the area contains +/-180 longitude.

I like both these ideas. We could have visibleCoordinateBounds.sw.longitude < visibleCoordinateBounds.ne.longitude for now. (That change would not be a breaking change because we don’t currently document the behavior when straddling the antimeridian.) When we’re ready for a major version bump, we can then change it to visibleCoordinateBounds.sw.longitude == +170.0 and visibleCoordinateBounds.ne.longitude == -170.0.

[pietrasm]

Sounds good, thanks a lot. I would really appreciate it. Just to clarify, this change will make it possible that visibleCoordinateBounds.sw.longitude > visibleCoordinateBounds.ne.longitude, not the other way around.