CI (Edit: attempt 2 testing composition demo_nodes_cpp image_tools pendulum_control lifecycle quality_of_service_demo_cpp quality_of_service_demo_py)

• Linux
• Linux-aarch64
• macOS
• Windows

@sloretz I don't understand why all these tests are timing out. I don't think it was from my changes. Any insight?

I don't understand why all these tests are timing out. I don't think it was from my changes. Any insight?

I believe those timeouts were introduced by ros2/launch#265 (which has since been reverted).

Let's try again:

• Linux
• Linux-aarch64
• macOS
• Windows

@rotu could you please rebase this PR.

@nuclearsandwich Done.

One last CI for luck.

• Linux
• Linux-aarch64
• macOS
• Windows

@rotu am I correct in expecting that we want to merge this before ament/ament_lint#148 ?

@nuclearsandwich, yes, so that CI doesn't break when we merge ament/ament_lint#148. I think it makes more sense to do it in the opposite order, so CI enforces correctness of this PR, but @dirk-thomas insisted on fixing the code before we fix the linter.

I think it makes more sense to do it in the opposite order, so CI enforces correctness of this PR,

I understand the logic of that workflow. But given that our team is so highly parallelized it doesn't make sense to break builds for the duration of a CI run. It's also possible to verify that the change resolves failures by triggering a CI run that includes ament/ament_lint#148 but not this change and then running the two together without affecting mainline CI for everyone else.

Edit: wrote the same sentence two ways and committed the merge conflicts...

That makes even more sense, and I didn't realize it was a possibility. How do I trigger or request such a CI build?

That makes even more sense, and I didn't realize it was a possibility. How do I trigger or request such a CI build?

The easiest thing to do is request CI on that ament_lint PR before this merges, then a second round after this does.