The conda build was also failing because it could not connect to the GEUS Dataverse. This is done in the unit testing for pypromice.get for fetching datasets. So I have commented these tests out to avoid this happening again.
I've also provided an option in the setup.py to offer the dependencies for pypromice's post-processing functionality (i.e. eccodes and scikit-learn) as a separate pip build. I.e.
pip install pypromice[postprocess]
I've not implemented this, but it is just an option that is currently commented out.
The conda distribution of pypromice was not building and I figured out it was because we had added eccodes as an explicit dependency. I.e.
This is now added in the conda build.
The conda build was also failing because it could not connect to the GEUS Dataverse. This is done in the unit testing for pypromice.get for fetching datasets. So I have commented these tests out to avoid this happening again.
I've also provided an option in the
setup.py
to offer the dependencies for pypromice's post-processing functionality (i.e. eccodes and scikit-learn) as a separate pip build. I.e.I've not implemented this, but it is just an option that is currently commented out.