Support non-anti-meridian split on alternate projections

[mourner]

For conical projections where center is close to the anti-meridian, such as Albers or Lambert for Alaska or New Zealand, having the divide across the anti-meridian is a significant user experience problem and could be a blocker for adoption.

We should find a way to move the split to the opposite side of the world from projection center so that it doesn't interfere with interaction around the target area. Not sure what's the best approach for this yet technically, but one idea is to use one more "world copy" on the left side of the anti-meridian, and modify tile clipping masks for tiles close to the longitude opposite from projection center.

[ansis]

Not sure what's the best approach for this yet technically, but one idea is to use one more "world copy" on the left side of the anti-meridian, and modify tile clipping masks for tiles close to the longitude opposite from projection center.

Sounds good to me. This same approach would let us put the center of the winkelTripel, etc, at a different longitude

[johnlaur]

Moving the "split" is equivalent to a pre-projection global offset (rotation) of longitude. I don't know enough about mapbox internals in this area to know if that makes it easier or not or how it relates to any future globe/3d surface work . Thanks again so much for the alternate projection work and for taking a look at this use case.

[tayalav]

This is a great point! I've just got some feedback from our users that's they couldn't use projections because we cut a small part of the continent on the east side

.

[stevage]

Thanks. I just came here to raise this issue. I'm based in Australia, and it's very noticeable that of the 6 projections now available, 4 of them put Australia at the edge of the world. I can't really see myself making much use of them for that reason (just confirming your suspicions about blockers to adoption).