Fix issue #9 - parameterization of dependency tests

[GalOz666]

Instead of using a helper function or a parameter - the original name of the node will be kept in the list of tests to check when using the depend=[] feature. If the function fails at one of its parameterized runs the node will remain in a failed state. Should there not be any parameterization there will not be any duplicate to check.

[GalOz666]

The failed test seems to be related to a change in Pytest between some versions where it will say "SKIP" and newer versions will say SKIPPED. Everything works fine - it seems to be more of an issue with the test then with the PR.

[RKrahl]

Yes, the issue with "SKIP" versus "SKIPPED" is a change in pytest 4.2.0 that causes test_03_skipmsgs.py to fail. I already did notice that, but I didn't got around to fix it yet. At least, I opened an issue now, see #34.

[GalOz666]

sounds good. I am still having issues with this so I will continue working and reopen this

[RKrahl]

Btw, I didn't had a detailed look into what you are trying to do in DependencyManager.addResult() and I'm afraid, I won't have the time to do so right now. But if it results in a semantic change in what is considered the dependency, something like depending on all parametrized test instances at once, I'm not sure I will be happy with that. I'd rather prefer a resolution of #9 by implementing the helper function sketched there.

[GalOz666]

Actually , my solution was to add the original name of the node to the results dictionary - such that you can reference it in depends and also reference the name of the nodes with parameters (eg., node[param, param] ) without interruption within the same module.

the only bug is that I need to fix (i think I just did) to make sure that one failure of the original-name node will STICK and never update as PASSED during the module test run.

[GalOz666]

If you are ok with this - I would write some tests for this in a future PR, with your permission of course.

[RKrahl]

Interesting idea. I have to think about that. As suggested, I'm somewhat busy right now, so I can't make promises when I will check that in detail.

[GalOz666]

No problem :) this is mainly for my own project but i think others will benefit as well. A bit hacky but it works fine now ;p

[GalOz666]

class TestPyDep:

@pytest.mark.dependency(name='smoke')
@pytest.mark.parametrize('nothing', [1, 1, 1])
def test_dep3(self, nothing):
    assert nothing

@pytest.mark.parametrize('nothing', [1, 1])
@pytest.mark.dependency(depends=["smoke"])
def test_two(self, nothing):
    assert nothing

@pytest.mark.parametrize('nothing', [1, 1])
@pytest.mark.dependency(depends=["TestPyDep::test_dep3"])
def test_three(self, nothing):
    assert nothing

^ first five tests (2 first tests) will run - last tests skipped.

class TestPyDep:

@pytest.mark.dependency(name='smoke')
@pytest.mark.parametrize('nothing', [1, 0, 1])
def test_dep3(self, nothing):
    assert nothing

@pytest.mark.parametrize('nothing', [1, 1])
@pytest.mark.dependency(depends=["smoke"])
def test_two(self, nothing):
    assert nothing

@pytest.mark.parametrize('nothing', [1, 1])
@pytest.mark.dependency(depends=["TestPyDep::test_dep3"])
def test_three(self, nothing):
    assert nothing

^ 1 failed, 2 passed, 4 skipped

[DevilXD]

Is this something that works? Is there a possibility it could be merged?

[GalOz666]

I am using it in production since posting it - and it does pass acceptance tests etc. feel free to try it out - you can find the forked version in my repositories

…

On Sun, 31 May 2020 at 22:30, DevilXD ***@***.***> wrote: Is this something that works? Is there a possibility it could be merged? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#33 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AJUW6AYRGXRTLJONMXV47WLRUKV45ANCNFSM4G43NMNA> .