nosedep causes all ancestor unites classes to be executed

[chickenandpork]

When testing #12, ran into another issue:

(based on decorated_functional_dep_skip, edited similar to #12, all depends removed, two descendent classes added)

from unittest import SkipTest, TestCase
  
class decorated_functional_dep_skip(TestCase):
    def test_dfds_a(self):
        pass

    def test_dfds_b(self):
        pass

    def test_dfds_c(self):
        pass

    def test_dfds_d(self):
        pass

    def test_dfds_e(self):
        pass

    def test_dfds_f(self):
        pass

class descendent1(decorated_functional_dep_skip):
    pass

class descendent2(descendent1):
    pass

Run with the following shows only 6 unittests as expected:

nosetests --no-byte-compile --with-xunit -v -s ./testfail.py:descendent1

Run with the --with-nosedep option, and now all ancestor and descendent classes (6 x 3 == 18 tests) also executed:

nosetests --no-byte-compile --with-nosedep --with-xunit -v -s ./testfail.py:descendent1

[Zitrax]

Right, I have always ran all the tests so specifying the class has not been tested.

But I can reproduce the issue and will have a look.

[Zitrax]

I have committed a fix for this as you can see above. I have not done much testing but it seem to solve this exact case though.

Can you check if it works for your use cases?

[chickenandpork]

Looks better on the surface; seems to give us an early bail-out when a dependency fails (using "@Depends(after=test_thing_I_depend_on)"). It *seems* to also work when the dependency(def test_thing_I_depend_on(self):) is (nose) parameterized as well, but our tests take time, it's difficult to run through all right away. So far, so good. (thanks, btw) Do you personally use parameterization ?

…

On Sat, Nov 17, 2018 at 5:08 AM Zitrax ***@***.***> wrote: I have committed a fix for this as you can see above. I have not done much testing but it seem to solve this exact case though. Can you check if it works for your use cases? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#13 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACNonCgGZBxXm7NsABggxn7JQmf6Wv3Iks5uwApBgaJpZM4YafEA> .

-- [email protected] "金鱼" http://linkedin.com/in/goldfish

[Zitrax]

Ok that sounds good. I'll hold off releasing for a while then to see if something else shows up.

I am not using parameterized tests with nose, so I am unsure of the status of that.