search

UniversalRobots / Universal_Robots_ROS2_Gazebo_Simulation

BSD 3-Clause "New" or "Revised" License
54 stars 24 forks source link

Ci setup #5

Closed fmauch closed 2 years ago

fmauch commented 2 years ago

This PR aims at improving the CI situation for this repo. Currently most builds are failing, but it is not necessarily obvious why and what is failing.

Therefore, I suggest to disable failing fast in binary and semi-binary builds, as there might be situations where the testing packages are fine, but the released ones are not or the other way around. When fail-fast ist enabled, the other job will get aborted before we see the result.

destogl commented 2 years ago

We need to update dependencies as they are in the driver. It seems that mostly driver dependencies are failing.

fmauch commented 2 years ago

We need to update dependencies as they are in the driver. It seems that mostly driver dependencies are failing.

For the previous runs there seemed to be a glitch in getting packages from the ubuntu sources. I restarted all failing tests and everything except for the semi-binary builds for galactic succeeded.

The galactic semi binary builds seem to fail due to missing support for initial_value. Which is absolutely fine, given that https://github.com/ros-controls/ros2_control/pull/635 is not yet merged. I would not suggest changing the dependencies in this repo, as we want to build it against the galactic branch in the end, right? I'd rather see the CI fail until the upstream blockers are merged.

I am wondering though, why the binary builds are fine. I guess, the initial_value urdf entries are simply ignored by the released versions of ros2_control and gazebo_ros2_control?

fmauch commented 2 years ago

Given that CI seems to behave as I would expect, I'll set this PR ready for review.

destogl commented 2 years ago

@fmauch rolling builds are currently failing because of "jammy" transition.