Rafael Schouten

Ok yes there is another bug with NamedTuple points your point is in Antarctica lol

Second bug is in GeoJSON.jl https://github.com/JuliaGeo/GeoJSON.jl/pull/57 After that you example is working for me Edit: I'm thinking we should make that `center` keyword accept a proper GeoInterace.jl point so you...

Sure, I agree. Note that that keyword is just a direct translation to the javascript, it's not something anyone really made a decision about.

Absolutely! But you should make another issue for that, as when I close this as completed whenever we can finally merge the GeoJSON.jl PR, your idea will be lost.

This really needs our `GeoInterface.convert(::Val{:GeoJSON}, geom)` method to not be a huge pain to write.

It still is better overall for simple `pois_rand` (and maybe others), as it allocates less than the alternatives. I get 10% improvement in a large simulation using `rand_pois` where `rand(Poisson(x))`...

Ok I see you're right, they have basically identical performance and no allocations. Distributions should swap the code comments for real documenation, those types are pretty much impossible to find...

I agree with that assessment @mkborregaard . Depending on GeoData.jl would be a larger change in philosophy and more like a decision to reduce the codebase here by half. GeoData...

Yep. GeoData.jl is not going anywhere if you want to cut some code and upgrade the tooling and functionality here later on. I just hope we can aim for integration...

> I suppose long-term the julian-esque would be to move all functions that rely on ArchGDAL to a shared lib and have any other heavy dependencies loaded from higher level...