Refactor rclcpp::executor::FutureReturnCode deprecation

[shivaang12]

---

Description of contribution in a few bullet points

• Remove deprecated rclcpp::executor::FutureReturnCode to rclcpp::FutureReturnCode

Future work that may be required in bullet points

[shivaang12]

I think this will have to wait till we shift to new ros2 nightly.

[SteveMacenski]

We are running nightly master builds in CI.

[shivaang12]

I've recently build ros2 from source and I cannot pass deprecation error without making this change. I am wondering why it's failing?

[SteveMacenski]

I did verify that that is a change between eloquent and master branch in rclcpp. I'm not exactly sure why that is, it takes a little while for the caches to update but that change looks like it happened about 2 weeks ago. It might be related to the Foxy freeze, maybe master builds have paused to be uploaded?

Can you update the PR name to be more descriptive. I think it might be hanging around for awhile until we sort it out

[SteveMacenski]

Lets spin this puppy again

[SteveMacenski]

Given other PRs built fine a few hours ago, I doubt anything has changed.

[SteveMacenski]

@shivaang12 try adding some extraneous info in ros2_dependencies.repos to break the cache. I believe that's what https://github.com/ros-planning/navigation2/blob/master/.circleci/config.yml#L203 is doing.

Add anything, doesn't really matter. A comment, a new package, etc

[shivaang12]

okay

[shivaang12]

Let me know if this will work

[SteveMacenski]

Cleaning Underlay
Input data is not valid yaml format: while parsing a block mapping
  in "<stdin>", line 2, column 3
expected <block end>, but found '<block mapping start>'
  in "<stdin>", line 22, column 4

well... it did something 🤖 but you added an extra space so it was invalid yaml in the ompl line

[shivaang12]

Oh, it was one space indent.

[SteveMacenski]

Wow, OMPL takes a long time to build. I guess I shouldn't be surprised with all those templates

[SteveMacenski]

I really hope after all of that building it doesnt fail from

�]0;colcon build [9/16 done] [4 ongoing]�Finished <<< image_common [2.89s]
�]0;colcon build [10/16 done] [3 ongoing]�[Processing: behaviortree_cpp_v3, gazebo_msgs, ompl]
--- stderr: behaviortree_cpp_v3
In file included from /opt/underlay_ws/src/BehaviorTree.CPP/src/action_node.cpp:166:0:
/opt/underlay_ws/src/BehaviorTree.CPP/3rdparty/coroutine/coroutine.h: In function ‘void coroutine::yield()’:
/opt/underlay_ws/src/BehaviorTree.CPP/3rdparty/coroutine/coroutine.h:367:11: warning: unused variable ‘stack_top’ [-Wunused-variable]
     char* stack_top = routine->stack + ordinator.stack_size;
           ^~~~~~~~~
/opt/underlay_ws/src/BehaviorTree.CPP/3rdparty/coroutine/coroutine.h:368:10: warning: unused variable ‘stack_bottom’ [-Wunused-variable]
     char stack_bottom = 0;
          ^~~~~~~~~~~~
/opt/underlay_ws/src/BehaviorTree.CPP/tools/bt_log_cat.cpp: In function ‘int main(int, char**)’:
/opt/underlay_ws/src/BehaviorTree.CPP/tools/bt_log_cat.cpp:27:10: warning: ignoring return value of ‘size_t fread(void*, size_t, size_t, FILE*)’, declared with attribute warn_unused_result [-Wunused-result]
     fread(buffer.data(), sizeof(char), length, file);
     ~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
---

[SteveMacenski]

Didn't work.

@ruffsl I think I know the answer but just need to verify - we added an extra package to break the cache and it rebuild the underlay workspace, that clearly didn't have a rebuild of ROS2 master. From looking at the results, it seems like ROS2 nightly comes from the docker image in stage 1 of circleCI.

That's generated from an automated build from dockerhub every time we push to master but since the BT.CPP and gazebo11 issue, those builds are failing so we're not seeing latest ROS2.

Is that correct?

Therefore steps to fix are

• Fix BT.CPP so your WIP PR works
• Fix gazebo 11 in your PR
• Let dockerhub rebuild an image when we push that to master
• rerun jobs

Correct?

[SteveMacenski]

@shivaang12 I think step 1 is to fix the issues with BT.CPP and update to the latest master branch per #1686 . Unfortunately, I'm not going to be in front of a computer for tonight (on phone). That would unblock Ruffin's PR that if it works would then probably stand everything else back up in a day or so.

[shivaang12]

Sounds good!

[SteveMacenski]

I keep going through this back and forth of forgetting how our CI is set up. I think it would be really nice if @ruffsl would write up a page for navigation.ros.org about how the CI system works. Some reference for people working with it (like me) that can't remember things apparently or new people that will eventually need to get up to speed about this because of the bus test. Even the basics would be great. What triggers what to happen, where Dockerhub meets CircleCI, where the caches are and how to break them, any changes needed between distributions, the role of each of the build steps and rationale for it, etc. Also may be helpful for other projects that may want to mirror our system.

[SteveMacenski]

@shivaang12 remove OMPL and rerun after these 2 build succeed

https://hub.docker.com/repository/registry-1.docker.io/rosplanning/navigation2/builds/b42c2d4c-625f-46a9-90df-bc240981afe9

https://hub.docker.com/repository/registry-1.docker.io/rosplanning/navigation2/builds/cc40b985-b302-4d5f-b0b5-492805bff625

It should work then

[SteveMacenski]

First build succeeded, second in progress

[shivaang12]

Somehow I can't see if the build has ended. Let me know the status so I will push the changes.

[SteveMacenski]

https://hub.docker.com/repository/registry-1.docker.io/rosplanning/navigation2/builds/cc40b985-b302-4d5f-b0b5-492805bff625

You can't see this link? It shouldn't require admin I don't think

[shivaang12]

Clicking the link takes me to homepage https://hub.docker.com/r/rosplanning/navigation2

[ruffsl]

@shivaang12 , please reminder to rebase onto current master to ensure your using the latest changes to the CI and codebase.

[SteveMacenski]

@shivaang12 its ready

[shivaang12]

I will do it in the morning, if that's okay?

[ruffsl]

From the CI, looks like this get the build step passing.

[SteveMacenski]

#1717

This has some test issues with cppcheck now, which is interesting