Misteryous test failing

[Juanlu001]

🐞 Problem

One particular test has been bugging us for a while now:

______________ src/poliastro/tests/tests_plotting/test_core.py _______________
142: error: "None" has no attribute "template"

It supposedly comes from here:

poliastro/src/poliastro/tests/tests_plotting/test_core.py

Lines 140 to 142 in 428db93

	def test_dark_theme():
	frame = OrbitPlotter3D(dark=True)
	assert frame._layout.template.layout.plot_bgcolor == "rgb(17,17,17)"

However, the test doesn't run if any filtering options are applied to pytest, therefore it's a pain in the neck to try to reproduce. I tried:

• TOXENV=py37 MPLBACKEND=Agg tox -- src/poliastro/tests/tests_plotting/
• TOXENV=py37 MPLBACKEND=Agg tox -- -k plotting

It started failing overnight, so I suspect it has to do with the MyPy version.

[Juanlu001]

• ✔️ August 19th https://circleci.com/gh/poliastro/poliastro/5148
• 🔴 August 26th https://circleci.com/gh/poliastro/poliastro/5152

[Juanlu001]

Differences in dependencies:

$ diff -u {ok,bad}_deps2.txt | colordiff
--- ok_deps2.txt	2019-09-14 13:31:16.377385995 +0200
+++ bad_deps2.txt	2019-09-14 13:31:25.561352437 +0200
@@ -38,7 +38,7 @@
 jsonschema==3.0.2
 jupyter-client==5.3.1
 jupyter-core==4.5.0
-keyring==19.0.2
+keyring==19.1.0
 kiwisolver==1.1.0
 llvmlite==0.29.0
 MarkupSafe==1.1.1
@@ -54,7 +54,7 @@
 numba==0.45.1
 numpy==1.17.0
 packaging==19.1
-pandas==0.25.0
+pandas==0.25.1
 pandocfilters==1.4.2
 parso==0.5.1
 pexpect==4.7.0
@@ -70,11 +70,11 @@
 Pygments==2.4.2
 pyparsing==2.4.2
 pyrsistent==0.15.4
-pytest==5.1.0
+pytest==5.1.1
 pytest-cov==2.5.1
 pytest-doctestplus==0.3.0
 pytest-mpl==0.10
-pytest-mypy==0.3.3
+pytest-mypy==0.4.0
 pytest-remotedata==0.3.2
 python-dateutil==2.8.0
 pytz==2019.2
@@ -103,8 +103,8 @@
 typed-ast==1.4.0
 typing-extensions==3.7.4
 urllib3==1.25.3
-virtualenv==16.7.3
+virtualenv==16.7.4
 w3lib==1.21.0
 wcwidth==0.1.7
 webencodings==0.5.1
-zipp==0.5.2
+zipp==0.6.0

Most likely offender is pytest-mypy.

[Juanlu001]

I confirm that with pytest-mypy this particular test passes. I will send a PR to pin the version and then try to isolate the failure.

[Juanlu001]

...I tried to narrow down the example, and I narrowed it down so much that pytest-mypy 0.3.3 fails with the same error. Which means that perhaps there is some error not being caught by the earlier version because of some code complexity.

[Juanlu001]

In fact, the issue seems to be that with pytest-mypy 0.3.3, that file gets ignored because of some missing import, but pytest-mypy 0.4 is somehow smarter, runs the check on the file, and realizes that there is something wrong. This is something to investigate on the mypy side.

[Juanlu001]

python/mypy#7513

Luckily, there is a workaround.

[Juanlu001]

Fixed in #778.

[Juanlu001]

This should be fixed in mypy 0.740 when it's available.