Pulls: #824
Gist: https://gist.githubusercontent.com/christophebedard/3a387055330bc86bf34c1c11a029c230/raw/3eefe972d23c9cd96fbc4243f027e4f8ba4e4d3d/ros2.repos
BUILD args: --packages-up-to launch
TEST args: --packages-select launch
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/15115

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

Yes I would say it is separate.

As in, the behaviour this is testing is correct?

If "correct" means "consistent with PYYAML 1.1", then yes I believe it's largely correct. I'm not an expert in the nuances of YAML specs, but from the discussion in ros2/launch_ros#384 (comment), I'm a bit on the fence on the _permute_assertion checks though.

If _permute_assertion intends to reflect YAML 1.1 / PYYAML, then there's a small inconsistency:

    _permute_assertion('1234e1', '12340', lc, 'true')  # Arguably wrong. PYYAML treats LHS as a string, so shouldn't equal RHS
    _permute_assertion('1234E2', '123400', lc, 'true')  # Arguably wrong. PYYAML treats LHS as a string, so shouldn't equal RHS
    _permute_assertion('1234E2', '1234E2', lc, 'true')  # OK, PYYAML treats the LHS as a string anyway
    _permute_assertion("'1234E2'", '123400', lc, 'false')  # OK, LHS is coerced to string
    _permute_assertion("'1234E2'", '1234E2', lc, 'false')  # OK, LHS is coerced to string
    _permute_assertion("'1234E2'", "'1234E2'", lc, 'true')  # OK I guess.

Though I have a feeling that, for original authors, the measure of "correctness" isn't YAML 1.1 but something else (maybe some heuristics to reflect how ROS 1 roslaunch behaved?). I should also mention that it's not yet clear to me what piece of code makes '1234e1' become equal to '12340'.

It's also interesting that '1234E1' == '12340' and '1234E1' == '1234E1' are simultaneously true.