Found some more aspects of testing that need improved:
- The
update_hourly subcommand isn't tested, so need to add something for that.
- Since the conda recipe only runs the fast tests, we don't catch errors in the packaging that only affect the longer tests. I'll need to set up some offline tests that try both a pip install and conda install (from a local channel + conda-forge) and run all the tests, checking every night if there are new commits in
main.
Found some more aspects of testing that need improved:
update_hourlysubcommand isn't tested, so need to add something for that.main.