raphael dussin

in #98 I propose to rename chunks='ND' by chunking_method='ND' not to be confused with the actual size of the chunk. Also CS needs a special treatment of the chunking_method

yes, I see some new failed builds now. I can take it from there.

got it! I am merging a new one with some fixes to the readthedocs.yml

merged commit and RTD doesn't build :(

no, saving changes in advanced settings usually triggers a build but it doesn't respond to following code changes

what did you do to fix it? I replaced the webhook but I didn't have a PR to merge to test if it was working.

The new internals allow for an arbitrary numbers of faces and padding if necessary. I have looked at the documentation here [cube sphere](http://mitgcm.org/public/r2_manual/final/online_documents/node233.html) and I think it might work out...

I tested the ocean part (global_ocean.cs32x15) with my latest code and it almost worked. The only issue is that the llc option (that allows the faces) assumes nx = ny...

I got confused with the number of tiles in the test case (12 tiles of 16x32)... I can make xmitgcm read the data but it doesn't look right. Is it...

Hi @brian-rose , sorry for the delay! I have been working on this today. The cube sphere output goes like this: (z,y,face,x). Although it is not straightforward to make it...