rework needed for "kill" subtest about autotest-docker

[cevich]

Thanks JC. For ref: some fixes (and some more trouble) is in #65

[cevich]

Moving this out to later version to give more time (0.7.5)

[ldoktor]

Hi @cevich, which one do you dislike? I know we have a different opinion on Exception, so I can remove them (I still believe cleanup should clean everything even on failure, but it's your choice).

About the old-style class it's just size optimization, I can add the (object) if you like

And the 'int' has no remove member is just false-negative. Based on the code it has to be list in order to get to the remove part.

[cevich]

For this and #65, I'm not interested in arguing the accuracy or usefulness of pylint or pep8 findings. I get that they're not perfect, and sometimes really annoying. The fact is, we use them simply to help enforce a consistent set of standards that generally lead to more reliable, predictable an easy to maintain code.

Our directive is for functional testing: Prove the existing, working parts of docker, remain so. In-depth and detailed/specific scenario testing is secondary to covering the basics. We really just want these tests passing over and over and over, simple and boring. Perhaps it would help to think about uncovering new/unknown bugs, as a side-effect rather than the central purpose.

Once this basic functional-test requirement is met, then work on extending (or adding additional) tests to cover more detailed, specific scenarios. Perhaps start by simplifying and dumbing-down the test (as much as possible) so pylint doesn't complain and most breakage points (here or in dockertest) would have obvious sources (for mid-level/junior engineers unfamiliar with the code). Then over time, work on pulling content back in bit-by-bit from the older commits, to cover the additional detail.

[cevich]

Okay, merged in via #201 Thanks for sticking with it, and putting up with my complaining :D