Duncan McGregor

For what it's worth, I can perform the same operations on NetCDF files with Xarray, so it doesn't seem a ridulous thing to do? ```python from dask.distributed import Client client...

Hi, thanks for the response. This is running in EC2 (Cloud9 fwiw), but my understanding of that `Client(n_workers=4)` line is that the Dask ‘cluster’ will just be on the Jupyter...

I’ve given @wjbenfold access to the repo, it seems we’re organisationally proximate!

I've tried a selection of files from https://github.com/SciTools/iris-test-data/tree/master/test_data/NetCDF, but can't find one that will open with `iris.load_cube`. If you can point me at a suitable candidate to run against the...

I'll have a chat with miniconda and see what I can do

I've tried again in a fresh and minimal 3.7 environment. ```bash conda create --name iris python=3.7 -y conda activate iris conda install distributed iris pytest -c conda-forge -y ``` Given...

We get the same issue on MacOS and a different MOGREPS-G NetCDF file.

Thanks for this. Loading into local memory should be possible for me, so happy to have this as just an irrititating edge-case.

Hey, sorry, wasn't notified of your issue for some reason. I can see that, for the value objects that Fakir is designed for, equality based on value would be a...

Is the special case much more convenient than just providing an implementation? ``` Map headers = ImmutableMap.of("Host", "www.example.com", ... ); new Faker() { String getHeader(String header) { return headers.get(header); }...