Backport PR #12549 on branch v5.0.x (TST: Temporarily skipping MRT write doctest)

[meeseeksmachine]

Backport PR #12549: TST: Temporarily skipping MRT write doctest

[pllim]

Wut...? See #12820

  File "run_astropy_tests.py", line 12, in <module>
    import pytest
  File "PyInstaller/loader/pyimod03_importers.py", line 495, in exec_module
  File "pytest/__init__.py", line 3, in <module>
  File "PyInstaller/loader/pyimod03_importers.py", line 495, in exec_module
  File "pytest/collect.py", line 8, in <module>
  File "PyInstaller/loader/pyimod03_importers.py", line 495, in exec_module
  File "_pytest/deprecated.py", line 13, in <module>
  File "PyInstaller/loader/pyimod03_importers.py", line 495, in exec_module
  File "_pytest/warning_types.py", line 8, in <module>
  File "PyInstaller/loader/pyimod03_importers.py", line 495, in exec_module
  File "_pytest/compat.py", line 36, in <module>
  File "py/_vendored_packages/apipkg/__init__.py", line 157, in __makeattr
  File "py/_vendored_packages/apipkg/__init__.py", line 75, in importobj
ModuleNotFoundError: No module named 'py._path'
[5162] Failed to execute script 'run_astropy_tests' due to unhandled exception!

[pllim]

@WilliamJamieson or @eslavich , what is going on here?

________________________ ERROR collecting test session _________________________
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/pluggy/_hooks.py:265: in __call__
    return self._hookexec(self.name, self.get_hookimpls(), kwargs, firstresult)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/pluggy/_manager.py:80: in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/pytest_asdf/plugin.py:318: in pytest_collect_file
    return AsdfSchemaFile.from_parent(
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/pytest_asdf/plugin.py:62: in from_parent
    result = super().from_parent(parent, fspath=fspath, **kwargs)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/_pytest/nodes.py:633: in from_parent
    return super().from_parent(parent=parent, fspath=fspath, path=path, **kw)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/_pytest/nodes.py:264: in from_parent
    return cls._create(parent=parent, **kw)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/_pytest/nodes.py:140: in _create
    return super().__call__(*k, **kw)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/_pytest/nodes.py:588: in __init__
    path = _imply_path(type(self), path, fspath=fspath)
../../.tox/py38-test-alldeps/lib/python3.8/site-packages/_pytest/nodes.py:110: in _imply_path
    warnings.warn(
E   pytest.PytestRemovedIn8Warning: The (fspath: py.path.local) argument to AsdfSchemaFile is deprecated. Please use the (path: pathlib.Path) argument instead.
E   See https://docs.pytest.org/en/latest/deprecations.html#fspath-argument-for-node-constructors-replaced-with-pathlib-path

[eslavich]

what is going on here?

Looks like pytest 7.0.0 was released and deprecates an argument that we use in AsdfSchemaFile's base class.

[pllim]

@eslavich , I just found asdf-format/asdf#1066 -- what is the timeline for a fix? 🙏

[WilliamJamieson]

@eslavich , I just found asdf-format/asdf#1066 -- what is the timeline for a fix? 🙏

I'm looking into it now.

[pllim]

I am just going to merge since we're fixing pytest compat in #12823 and upstream.

[dhomeier]

OK I'll take your word for it – I am a bit confused why the devdeps test that was already running with asdf-format/asdf#1075 is now marked as cancelled, but obviously the fixes from #12823 are needed as well.

[pllim]

Oh, I cancelled it. I didn't know you restarted it. Sorry!

[dhomeier]

I decided to re-run them once the asdf fix was merged, but had not realised it now takes those tests 3 hours to run!

[pllim]

Anyways, devdeps is running on the merge commit in v5.0.x here: https://github.com/astropy/astropy/runs/5073451955?check_suite_focus=true

Hopefully it will pass, or at least fail with a different error. 🤞

[pllim]

I cancelled the devdeps job again. It is not supposed to take this long and it looks like it is going to get killed for taking too long anyway. Something is wrong but let's put out the pytest 7 fire first.