Replace ready_fn with ReadyToTest action#391
Conversation
|
I don't understand these failures, but some of them are the same on Linux and macOS. WIll investigate: In this particular test I only swapped out the ReadyToTest action for the ready_fn - maybe there's some unintended difference about the timing that I need to understand better |
Signed-off-by: Pete Baughman <pete.baughman@apex.ai>
a08c4fe to
f2047d2
Compare
|
@sloretz Sorry this has been sitting here forever - I've just started working on this again and I've got it all rebased. I'm not seeing the same failure locally - do you mind kicking off another CI job? It's possible a change in launch_testing in the last 3 months has fixed this. |
|
Thank you - the error has changed a little, but the root cause appears to be the same. I will continue to investigate |
|
I see now. . . this is a pretty wild test-within-a-test. The process that's hanging is called And then hangs forever, apparently. I can get the issue to happen on my machine now - unfortunately it also happens on 'master' for me so I'm not sure how I can proceed with this PR. @jacobperron it looks like this is hanging here If while it's hung, I do ros2 service list from another terminal, I see: So it looks like the service that it's trying to call is there. Do you have any insight into why this might be hanging? I'm getting the hang by running |
|
Sounds like a occurrence of ros2/build_farmer#198. I don't have any ideas at the moment. |
|
@jacobperron Wild - I can get it to happen every time on my machine - let me know if there's anything you think of to try and I'm happy to dig deeper. |
|
Any idea how I should proceed with this? It seems like I'm blocked on some unrelated bug. |
|
Looking at the lastest CI run on this thread, I looks like all of the macOS failures are occurring on our nightly CI, but the |
|
At least on my machine, it appears that the test is failing because the parameter server really is hanging. I can improve the output of the test a bit to make it easier to see what's happening, but as far as I can tell it's not a problem with the test |
|
The failures present in the latest CI I've triggered are also occurring on our nightlies, so I'd say that this change is good to merge. |
The ready_fn will be deprecated in the future in favor of the ReadyToTest() action in launch_testing. See ros2/launch#346 (comment) for background information
Note I had some trouble getting these tests to build and run locally, so I'll pay extra-close attention to the CI jobs on this one
Signed-off-by: Pete Baughman pete.baughman@apex.ai