Alexander Barth

I am closing this issue. Feel free to re-open should the issue persists. Thank you for your time to make this issue easily reproducible!

Thank you for your insightful comments! Indeed, ` Statistics.mean` (and friends) are not overloaded and thus they are fetching every element individually from the netCDF file which is slow. I...

I had a look at `HDF5.jl` (https://github.com/JuliaIO/HDF5.jl/blob/master/doc/hdf5.md#reading-and-writing-data) and they use `read`, for example: ```julia variable = read(ds["myvar"]) ```

Can you paste here the output of "ncdump -h file.nc" ?

Maybe the netcdf file uses some unimplemented type. Currently, only vlen-arrays as user types are implemented. https://www.unidata.ucar.edu/software/netcdf/docs/group__user__types.html#gaf4340ce9486b1b38e853d75ed23303da Can you retry with the updated code from master? It probably still fails,...

I have to say, that I am not knowledgeable at all on this topic. The EPSG codes, WKT and the [grid mapping attributes]( https://cfconventions.org/Data/cf-conventions/cf-conventions-1.9/cf-conventions.html#appendix-grid-mappings) are all redundant way to express...

Can somebody prepare a PR?

Indeed, `ncdump` (netcdf-4.7.3, OS: Linux, Ubuntu 20.04) has the same issue but it does work with netcdf-4.8.0 (at least on Linux). Currently Windows is blocking the upgrade to a newer...

If a windows use could test netcdf-c with the mingw compiler (available e.g. in [MSYS2](https://www.msys2.org/) or cygwin) and report back to Unidata/netcdf-c any issues when running the commands, that would...

Interestingly, NCDatasets on Linux can handle such files: ``` julia> using NCDatasets; NCDataset("Ω.nc") NCDataset: Ω.nc Group: / ``` So it seems that on Linux, the problem is limited to Ubuntu's...