Test Helm on different shells about acceptance-testing HOT 8 CLOSED

@jdolitsky Good idea. And it turns out to be much easier than adding a new target.

from acceptance-testing.

Do you mean "Test Helm shell completion across different shells..." or did you have something larger in mind?

I think the shell matters greatly for completion but I don't believe it will affect other features.

from acceptance-testing.

@marckhouzam I think we should just add a shells.robot (or something similar), with the first test case inside being Verify command completion works. This opens up the possibility of adding other shell-specifc tests in the future.

Please see kubernetes_versions.robot#L45 for an example (similar to Verify --wait flag works as expected test case)

from acceptance-testing.

@jdolitsky Sounds good. I'll give it a try.

from acceptance-testing.

@jdolitsky The way things are currently setup, when I add a shells.robot file, then whenever I want to run the shell-completion tests, it will run the entire acceptance test-suite, which is already long (over 7 min on my machine). So to test completion during development, this becomes difficult.

Would it be ok to split things so I could do make shells which will only run shells.robot?
And make acceptance can run both shells.robot and kubernetes_versions.robot?

from acceptance-testing.

@marckhouzam It would be my preference/suggestion that instead of adding a separate make target, we find the appropriate way to do this with robot. Then we can maybe have an option ROBOT_RUN_TESTS that defaults to all tests ( ROBOT_RUN_TESTS=kubernetes_versions,shells).

This way you could do something like

ROBOT_RUN_TESTS=shells
make acceptance

from acceptance-testing.

Proposed implementation of ROBOT_RUN_TESTS in #10

from acceptance-testing.

closed via #12

from acceptance-testing.