Fabian Gans

> Yeah maybe. I'm wondering if @meggart had a reason for always using Chunked() No, I did not know that there is a `TILED=NO` and I just assumed GDAL datasets...

> Probably Base.PermutedDimsArray should be forwarded to the same (like your proposed fix, but in DiskArrays). This proposed fix does not work because the `PermutedDiskArray` already wraps a PermuteDimsArray, and...

Thanks @AStupidBear If you have any questions or you stumble over issues with DiskArrays, feel free to open an issue.

I think there is a feature in NetCDF.jl that is not present in this package. The only and most important reason for me to use NetCDF.jl instead of this package...

Regarding the future for these packages, we might think about reaching out to JuliaIO. There are NetCDF files in the wild which are nit in the Geo domain (like chemistry)...

Just wanted to mentioned that I opened [this issue](https://github.com/JuliaIO/Roadmap/issues/4) at JuliaIO requesting to move NetCDF.jl there. Just in case someone wants to add something.

> Will you still want to do that @meggart? Yes, regarding pure maintenance of the package I would like to continue, because there is some legacy code of mine that...

> An essential part of the multifile code (with an aggregation of a given dimension) is implemented in a submodule CatArrays which does not import anything from NCDatasets This is...

> Do I understand it correctly, that the disk access will be done in chunks of data? Yes > It seems that getchunksize / haschunks are the chunk size /...

And just to add one thing. The chunks are only used for broadcast and reductions. Simple calls to `getindex` and `setindex!` will directly be translated to a `readblock!` call.