I really like this idea, fits well with our geospatial focus and the samples look really cool.

But shouldn't this be an example using a customized/subclassed ArcLayer?

To me this seem like an odd feature to add to a generic layer, especially a layer that currently works in any coordinate system, not just geospatial.

Specifically, I see a lot of extra lng/lat based shader code and a new prop that only apply to a very specialized case. I am assuming that 95% of ArcLayer users won't use this?

Changes made to ArcLayer are reverted. GreatCircleLayer is created as a subclass of ArcLayer in the experiment layer folder. A GreatCircleLayer example is added to layer browser(Experimental Core Layers) as can be seen below:

Changes made to ArcLayer are reverted. GreatCircleLayer is created as a subclass of ArcLayer in the experiment layer folder. A GreatCircleLayer example is added to layer browser(Experimental Core Layers) as can be seen below:

Thanks, that makes a lot more sense.

I could argue that this might be more of an "example specific layer" rather than an "experimental layer" (i.e. users would have to copy it from an example rather than import it from experimental layers), but I am fine with either approach.

Having it in experimental layers makes sense if we think there is a future layer catalog where this layer could fit in, I could for instance imagine a layer catalog with various geospatial "helper layers", other layers could perhaps a "compass layer", ...

Also, you could probably build some really cool geospatial grid layer/satellite orbit layers using this layer as a primitive.

Please update the PR title and summary to reflect the latest changes.