Rafael Schouten

@yeesian I agree, regular arguments should be spatial dimensions. M could use a keyword argument.

Yeah, that would work. Otherwise just use a different function with m last.

Defining `getproperties` for the row object should mean the `rowaccess`-based Tables.jl interface will just work too? Thats how a Vector of `NamedTuple` works. Its also nice to have that symmetry...

I was thinking about this yesterday for ArchGDAL. Better type stability should also help with compilation time, which could also be profiled at the same time.

Do you mean check if the return value is a band, and if so dont close it? Not allowing return values would break a lot if things.

This may be an issue with the difference between GDALs "natural" and actual block sizes.

Yeah maybe. I'm wondering if @meggart had a reason for always using Chunked()

This would be great, I've had a similar response to @visr a few times trying to work out how to pass options. Keyword arguments feel the most Julian, but a...

Yeah it's probably possible to allow both.

No problem! The above example mostly works, we should remove the GeoArray parts and put it in the docs. But its not a complete example yet - `setgeotransform` doesn't work....