deprecate the static single threaded executor#2598
Conversation
Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
clalancette
left a comment
There was a problem hiding this comment.
I'm totally on board with this deprecation, however:
- We are almost certainly going to have to make changes to the tests to either remove tests from StaticSingleThreadedExecutor, or suppress this warning (CI should show this).
- I'd like to get @wjwwood and/or @mjcarroll to approve.
mjcarroll
left a comment
There was a problem hiding this comment.
I approve (and this was always the goal).
We should probably suppress the warning for this iteration before removing entirely next time (to stick with our tick-tock policy) |
fujitatomoya
left a comment
There was a problem hiding this comment.
lgtm with green CI
… tests Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
|
I removed the performance benchmark tests and suppressed the deprecation warning for the unit-tests. |
clalancette
left a comment
There was a problem hiding this comment.
It looks like there are some uncrustify fixes that are needed, but otherwise this looks good to me.
Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
|
Pulls: #2598 |
i created an alias to give the deprecated executor a new name; this works when the class is directly used but it doesn't work in combination with templates (like std::make_shared<DeprecatedAlias>) because the compiler needs to resolved the type behind the alias triggering the warning Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
|
Ok, I can see the windows warnings here https://ci.ros2.org/job/ci_windows/22373/msbuild/#issuesContent. New CI: |
|
This pull request has been mentioned on ROS Discourse. There might be relevant details there: https://discourse.ros.org/t/next-client-library-wg-meeting-friday-16th-august-2024/39130/1 |
|
So this generally looks good to me now, but the Rpr job has warnings related to deprecations of the |
Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
Signed-off-by: Alberto Soragna <alberto.soragna@gmail.com>
|
@clalancette The warnings in the Rpr job were caused by a new test that was recently introduced and that was not using the I merged with Here a new CI |
|
I'm going to merge this since everything is green now! |
|
This pull request has been mentioned on ROS Discourse. There might be relevant details there: |
This PR marks the
StaticSingleThreadedExecutoras deprecated.With the current implementation, all other single threaded executors have better performance than this one and they are more stable.
See https://discourse.ros.org/t/the-ros-2-c-executors/38296 for a recent performance comparison that I run.