working on RunTimeWarning in tests

[sbillinge]

first clean up applycutoff method to make its purpose clearer and to bring it closer to pep8 compliance.

then figure out how to work on the test warning

[sbillinge]

@bobleesj @cadenmyers13 here is a PR which would be my first cut at addressing this warning.

First I try and understand what the function is supposed to be doing. It is basically creating a big array for the reciprocal space and then setting the voxels outside the measured Q-range to nan.

This is probably the desired behavior because they will always stay NaN. They may not stay zero if we set them to zero.

[sbillinge]

I think the test is probably at fault then. It is a warning that is being thrown by numpy for some reason. We may want to try and catch that warning or handle it somehow in the test.

[sbillinge]

we may be able to handle this by calling the test in a context manager, something like with self.assertWarns(RuntimeWarning):

[sbillinge]

OK, so actually it is being thrown by another method in fourigui, intensity_upd_global() so we may want to figure out what that is doing and if we need to throw a warning. When we do that we can take the opportunity to turn those strings into f-strings....

For now I have to work on something else with the experiment, but maybe you guys can get the idea from what I did here and take it forward.... Let me know if you want me to merge these changes, which don't fix the problem but might be useful in your fix

[sbillinge]

see inline

[sbillinge]

I wonder if these should be < and > rather than <= and >= since it is tagging the unwanted regions as true. i.e., presumably Qmax is a valid value not an invalid value so we don't want to set Qmax to zero.

[sbillinge]

Since we are moving away from unittest and to pytest, maybe we want to just junk this test and replace it with a new test that tests the behavior that we want?

[sbillinge]

ok, a bit more digging and it is numpy.nanmax that returns the runtimewarning but this is desired behavior as we expect some slices to have all nan's given our function. So we would just like to figure out how to suppress this warning polluting our output. We could also just ignore it, but future selves will end up revisiting this again I am guessing, so if we can suppress it in the test, this would be desirable. I think we could also catch it and suppress it in the code as a, possibly better, alternative.

[sbillinge]

Do you guys want to look into that?

[bobleesj]

Will wait for @cadenmyers13 to first give his thought since it's his cookie

[cadenmyers13]

@sbillinge Merging this will be fine, then I'll refer back to the comments. Thanks!