RMSD: added keyword superposition

[orbeckstASU]

FIxes #25

Changes made in this Pull Request:

• RMSD is now doing a RMSD-superposition by default
• disable with superposition=False

PR Checklist

• [x] Tests?
• [x] Docs?
• [x] CHANGELOG updated?
• [x] Issue raised/referenced?

[orbeckst]

@kain88-de any idea why the tests died? – It appears that we fail to properly start distributed.Client() https://travis-ci.org/MDAnalysis/pmda/jobs/359726635#L2039

Minor thing: How to tell pytest to only gather the test_*.py files and not also the normal code? https://travis-ci.org/MDAnalysis/pmda/jobs/359726635#L1994 – or is it doing something else??

[kain88-de]

Minor thing: How to tell pytest to only gather the test_*.py files

What you are seeing is the PEP8 test. They only check the formatting of the files. It's best to enforce good code style from the beginning.

I don't know where the timeout came from. Do the tests pass locally?

[orbeckst]

The tests pass locally.

Will check PEP8, thanks.

[orbeckst]

The line https://travis-ci.org/MDAnalysis/pmda/jobs/359726637#L2079

self = LocalCluster('tcp://127.0.0.1:44955', workers=0, ncores=0)

looks suspicious to me: does 0 workers and 0 cores mean "take all of them"? Is there a way to just get two workers with 1 process each with the simple setup or do we need to explicitly start scheduler and workers external of the tests?

[orbeckst]

Also, the PEP8 (?) tests say

pmda/rms.py FAILED

but I don't see any indication what is wrong. The pylint job looks fine.

[kain88-de]

Also, the PEP8 (?) tests say

pmda/rms.py FAILED

last line of the log. Pylint has nothing to do with the pep8 check.

[kain88-de]

The failure happens because of dask update. If I rerun a master branch build that succeeded I get the same error you have. Unfortunately I don't have time into looking for a proper fix.

[orbeckstASU]

Thanks for all the comments!

-- Oliver Beckstein Assistant Professor of Physics Arizona State University email: oliver.beckstein@asu.edu web: http://becksteinlab.physics.asu.edu

Am Mar 30, 2018 um 06:28 schrieb Max Linke notifications@github.com:

The failure happens because of dask update. If I rerun a master branch build that succeeded I get the same error you have. Unfortunately I don't have time into looking for a proper fix.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[orbeckst]

Need to fix #30 for the tests to make sense.

[orbeckst]

Apparently it does not work to decorate fixtures with xfail so I added the decorator to the tests and amended the history.

[orbeckst]

I don't understand why travis still fails, even with the xfails in place.

[orbeckst]

When I try the test locally in a conda env with MDA 0.17.0 I get

(pmda) yngvi:pmda oliver$ pytest -v pmda/test/test_rms.py
============================= test session starts ==============================
platform darwin -- Python 3.6.5, pytest-3.5.0, py-1.5.3, pluggy-0.6.0 -- /Users/oliver/anaconda3/envs/pmda/bin/python
cachedir: .pytest_cache
rootdir: /Users/oliver/MDAnalysis/pmda, inifile:
plugins: xdist-1.22.2, forked-0.2
collected 3 items

pmda/test/test_rms.py::TestRMSD::test_rmsd PASSED                        [ 33%]
pmda/test/test_rms.py::TestRMSD::test_rmsd_step xfail                    [ 66%]
pmda/test/test_rms.py::TestRMSD::test_rmsd_single_frame xfail            [100%]

===================== 1 passed, 2 xfailed in 1.66 seconds ======================

so the xfail works locally.

[orbeckst]

I am restarting the build ... just because I have no idea what to do.

[orbeckst]

@kain88-de any idea why the xfail decorator is not working on travis? https://travis-ci.org/MDAnalysis/pmda/jobs/364000696#L2064

[kain88-de]

did you compare pytest versions

[orbeckst]

Same versions:

• on travis Py 3.5: pytest-3.5.0
• Locally: pytest-3.5.0

[kain88-de]

we install a development build of mdanalysis that identifies correctly but doesn't have the DCD fix applied yet. So Travis has version 0.18.0 which doesn't fail.

[orbeckst]

I think I understand the problem: Travis uses the MDA 0.17.1-dev from @kain88-de 's channel and I don't think that those builds already have the fix for MDAnalysis/mdanalysis#1819 (because they were updated last time 2 months ago). BUT: these builds are versioned 0.17.1-dev and so the xfail considers them as tests that should pass.

My local tests passed with MDAnalysis 0.17.0 and the xfail behaved properly. Should I switch pmda tests to using the the standard MDA distribution or is there another reason that we are using the develop builds?

[orbeckst]

I came to the same exclusion but took 3 minutes longer to post...

[codecov-io]

Codecov Report

Merging #28 into master will increase coverage by 4.9%. The diff coverage is 100%.

@@            Coverage Diff            @@
##           master      #28     +/-   ##
=========================================
+ Coverage   93.37%   98.27%   +4.9%     
=========================================
  Files           7        8      +1     
  Lines         317      348     +31     
=========================================
+ Hits          296      342     +46     
+ Misses         21        6     -15

Impacted Files	Coverage Δ
pmda/test/test_rms.py	100% <100%> (ø)
pmda/rms.py	100% <100%> (+100%)	:arrow_up:
pmda/test/test_contacts.py	97.67% <100%> (-1.17%)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 69e84d2...78aa84a. Read the comment docs.

[kain88-de]

tests pass now. I also merged now that we test against the stable version of mdanalysis