Alexander Barth

I would (by far) favour that the order of dimensions is lon, lat, time in Julia if the file is written as time, lat, lon in python (when written with...

Thanks a lot for your time looking at this. I agree that this would be better in a separate package. Unfortunately, I am not enough "invested" in the GeoTIFF ecosystem...

> In Zarr the FillValue does not necessarily mean that it is missing. It is the default value that you would get when a chunk is not written. This is...

In my example, you do not need to modify the data on disk. `fillvalue` is a keyword argument of the function `cfvariable`. Setting it to nothing, makes NCDatasets to ignore...

In this commit https://github.com/JuliaGeo/ZarrDatasets.jl/commit/a84ef0ae0d5d4d8e2036d763a2375c8256ef0da8 ZarrDatasets no longer considers Zarr fill_value the same as the CF _FillValue for coordinate variables. I think that this should help with the present issue. (This...

I use Zarr in a generic non-interactive library, where the user can download datasets from different sources to prepare the initial conditions and boundary conditions for the geophysical ocean model....

I like also the current behavior of `Zarr.jl` (reversing the python dimension lat/lon into lon/lat, and not transposing the data), because this is also the current behavior of reading NetCDF...

I can confirm that this PR is still necessary for `MetadataCredentials()`. Would it be possible of somebody from the JuliaCloud org would merge this PR? Thanks for this great package...

@EhsanMehdipour can you prepare me an example data (similar to ds_test.nc) where we have for example the variable `DIATO_uncertainty` ?

maybe this ? ```julia ds_.attrib["auxdata_filenames"] = getindex.(auxdata_files,:filename) ds_.attrib["auxdata_varnames"] = getindex.(auxdata_files,:varname) ... ```