JUnit XML output: Include timezone information in XML

[wmanley]

JUnit writes local-time without a timezone by default. This means that
JUnit XML parsers default to local-time. The portal doesn't know what
timezone the user of stbt-rig wants so these timestamps can be interpreted
incorrectly. This change includes timezone information in the XML output.

There is a small chance of incompatibility caused by this change. If you
experience this please get in contact at support@stb-tester.com.

TODO:

• Test against real portal
• Test with real Jenkins

[wmanley]

Hmm, CI failed with:

AttributeError: module 'isort' has no attribute 'SortImports'

Looks like a version locking issue.

[drothlis]

Well, it looks like AppVeyor has caught an incompatibility with new versions of pytest:
https://ci.appveyor.com/project/drothlis/stbt-rig/builds/35072163/job/atimtgkwgppxebqj

________________________ ERROR collecting test session ________________________
c:\python36-x64\lib\site-packages\pluggy\hooks.py:286: in __call__
    return self._hookexec(self, self.get_hookimpls(), kwargs)
c:\python36-x64\lib\site-packages\pluggy\manager.py:93: in _hookexec
    return self._inner_hookexec(hook, methods, kwargs)
c:\python36-x64\lib\site-packages\pluggy\manager.py:87: in <lambda>
    firstresult=hook.spec.opts.get("firstresult") if hook.spec else False,
C:\projects\stbt-rig\stbt_rig.py:1227: in pytest_collect_file
    return StbtCollector(path, parent)
c:\python36-x64\lib\site-packages\_pytest\nodes.py:95: in __call__
    warnings.warn(NODE_USE_FROM_PARENT.format(name=self.__name__), stacklevel=2)
E   pytest.PytestDeprecationWarning: Direct construction of StbtCollector has been deprecated, please use StbtCollector.from_parent.
E   See https://docs.pytest.org/en/stable/deprecations.html#node-construction-changed-to-node-from-parent for more details.

I'm going to stop working on this now so back over to you @wmanley.

[wmanley]

Reproduced locally with:

$ python3 -m venv new-pytest
$ source new-pytest/bin/activate
$ pip install -r requirements.txt
...
$ pytest --version
pytest 6.0.1
$ pytest *.py
...

[wmanley]

Now there are only pylint issues remaining:

************* Module stbt_rig
stbt_rig.py:1237:4: [W0223(abstract-method), StbtCollector] Method 'get_closest_marker' is abstract in class 'Node' but is not overridden
stbt_rig.py:1237:4: [W0223(abstract-method), StbtCollector] Method 'gethookproxy' is abstract in class 'FSCollector' but is not overridden
stbt_rig.py:1237:4: [W0223(abstract-method), StbtCollector] Method 'isinitpath' is abstract in class 'FSCollector' but is not overridden
stbt_rig.py:1257:4: [W0223(abstract-method), StbtRemoteTest] Method 'get_closest_marker' is abstract in class 'Node' but is not overridden

This is a curious one. I think the get_closest_marker issue is casued by a limitation of pylint caused by a curious pattern in pytest.

Source:

class Node(metaclass=NodeMeta):
    ...
    @overload
    def get_closest_marker(self, name: str) -> Optional[Mark]:
        raise NotImplementedError()

    @overload  # noqa: F811
    def get_closest_marker(self, name: str, default: Mark) -> Mark:  # noqa: F811
        raise NotImplementedError()

    def get_closest_marker(  # noqa: F811
        self, name: str, default: Optional[Mark] = None
    ) -> Optional[Mark]:
        """return the first marker matching the name, from closest (for example function) to farther level (for example
        module level).

        :param default: fallback return value of no marker was found
        :param name: name to filter by
        """
        return next(self.iter_markers(name=name), default)

So it's repeated in the function definition. The last definition will take precedence. Not sure why they've done this.

[wmanley]

Regarding gethookproxy and isinitpath, the same issue is reported here: pytest-dev/pytest#7591 and will be fixed in 6.1: pytest-dev/pytest#7648.

I guess I'll just have to disable this lint :(