Closed jbusecke closed 2 years ago
@paigem if you think those changes are sufficient, I think we can merge this after the tests pass.
I think once this is done, we could release either 0.3.0 (since this has quite a few major new features) or 0.2.6. Maybe @paigem wants to do the release?
Ooh I don't even know where to start in making a release! But maybe that would be a good learning opportunity.
Also, the CI tests didn't all pass. Looks like the test_range_check_nan()
is raising an AssertionError
from line 80 for each variable. @jbusecke
Ooh I don't even know where to start in making a release! But maybe that would be a good learning opportunity.
That would be the point 😜. Also its pretty easy (if things dont break haha).
Thanks for fixing the CI!
I have implemented a checking function on the numpy level that can be turned on/off. This function checks the 'shrunken' arguments for nans and values that are out of the allowed range. This should completely eliminate the need for propagating errors from the fortran level (#4, #11), since we are now catching eveything that would lead to a
STOP
in fortran at the python level.