Pytest Style: auxiliary/base.py

[utkbansal]

Fixes #

Changes made in this Pull Request:

PR Checklist

• Tests?
• Docs?
• CHANGELOG updated?
• Issue raised/referenced?

[kain88-de]

please use the pytest.raises context manager

[utkbansal]

That's odd. The tests pass on my system.

[utkbansal]

@kain88-de @richardjgowers @jbarnoud base.py isn't collected for tests and hence test_get_bad_auxreader_format_raises_ValueError isn't executed. Should I move it to test_xvg.py or should we rename base.py?

[utkbansal]

@kain88-de Is it normal that I have to uninstall and reinstall MDAnalysis & MDAnalysisTests to reflect the changes I make in the tests?

[utkbansal]

@kain88-de @richardjgowers @jbarnoud On running test_iterate_as_auxiliary_from_trajectory I get the error ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() The error seems logical, how is it supposed to even compare it, there is ambiguity. How should I fix this?

[mnmelo]

That's odd, but I guess it depends on how exactly you're installing/running. Checking MDAnalysisTests.__path__ might also shed light on why you're not getting the behavior you think you should be getting.

[utkbansal]

@mnmelo This is what I see.

>>> MDAnalysisTests.__path__
['MDAnalysisTests']

I have setup using the following commands

pip install -e package/
pip install -e testsuite/

[mnmelo]

Assuming you're in your develop tree, then that seems ok, and you shouldn't need reinstalls. What exactly is the symptom that prompts you to reinstall MDAnalysis?

[jbarnoud]

On 07/16/2017 11:06 AM, mnmelo wrote: Assuming you're in your develop tree, then that seems ok, and you shouldn't need reinstalls. What exactly is the symptom that prompts you to reinstall MDAnalysis? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1492 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABUWujr2CkM-7Ef4VhDxWEh6CAOW2Nhqks5sOdKEgaJpZM4OZHV6>.

I recently reinstalled the package and the testsuite because I had failure with the new tests for libmdaxdr. I only had to do it once, though. I also had to do it when the new DCD reader was merged, but I only reinstalled the package at that time, and it worked fine.

[mnmelo]

Ah yes, a reinstall will be needed (once) if you pull any changes to MDAnalysis that involve changes to the C/Cython parts of the code. Was that the case, @utkbansal?

[mnmelo]

As to the test failure, there's a == comparison between arrays, which returns a bool array that can't be converted to a single bool value. I can't readily see the code and contents of the aux data being compared, but it seems either the wrong data are being compared or not being frame-indexed properly.

[utkbansal]

What exactly is the symptom that prompts you to reinstall MDAnalysis?

The basic symptom is that I make changes to tests and there is no effect when I run them. And that sometimes(like in this PR) causes the Travis build to fail though tests on my local system pass.

Ah yes, a reinstall will be needed (once) if you pull any changes to MDAnalysis that involve changes to the C/Cython parts of the code. Was that the case, @utkbansal?

This might have been the problem, but in this PR I haven't touched any .pyx files.

[kain88-de]

@utkbansal I recommend you work in a virtual env or conda-env. When you use your global env and one time installed without the -e flag you might see such problems.

[utkbansal]

@kain88-de I'm already using a virtualenv.

[mnmelo]

As for the failing array comparisons: you blindly replaced all assert_equal by assert. The two are not equivalent: numpy's assert_equal takes care of comparing arrays element-wise (it was also decided in #1444 to keep the more complete asserts from numpy, except assert_). You should roll back those changes, except maybe for cases where they were being used to compare single values.