Deprecate StaticSingleThreaded Executor. #4812
Conversation
Signed-off-by: Tomoya Fujita <Tomoya.Fujita@sony.com>
|
@clalancette @alsora could you review this? |
|
HTML artifacts: https://github.com/ros2/ros2_documentation/actions/runs/11670308253/artifacts/2142625163. To view the resulting site:
|
|
|
||
| The *Static Single-Threaded Executor* has been deprecated, and *Single-Threaded Executor* is recommended instead. | ||
| The *Static Single-Threaded Executor* was the only *Executor* optimized the runtime costs for rebuilding the entities of a node in terms of subscriptions, timers, service servers, action servers, etc. | ||
| But this runtime improvements are now available in all the other *Executor*, so that all *Executor* can now take advantage of this improvement. |
There was a problem hiding this comment.
I think the two sentences here can be written in a more clear way.
The Static Single-Threaded Executor was the only Executor optimized the runtime costs for rebuilding the entities of a node in terms of subscriptions, timers, service servers, action servers, etc.
But this runtime improvements are now available in all the other Executor, so that all Executor can now take advantage of this improvement.
could become
The Static Single-Threaded Executor was developed to reduce the the runtime costs for scanning the entities of a node in terms of subscriptions, timers, service servers, action servers, etc.
These runtime improvements are now available also in all the other Executor.
There was a problem hiding this comment.
We could also mention that there are known bugs affecting only the static executor
There was a problem hiding this comment.
@alsora do you mean ros2/rclcpp#2589, right? i am not sure about that, because that looks like regression and we are going to fix that? if it seems to take much longer time to address it, probably we can consider that with another PR?
There was a problem hiding this comment.
@alsora friendly ping.
There was a problem hiding this comment.
No, that ticket refers to a bug affecting all waitset-executors, but there are also some bugs that are specific to the static executor.
You can see them looking at the unit-tests that are run on StandardExecutors (i.e. they are not run on the static executor)
Signed-off-by: Tomoya.Fujita <Tomoya.Fujita@sony.com>
There was a problem hiding this comment.
This looks good to me, but I'd like to get @alsora 's approval before we merge.
Signed-off-by: Tomoya Fujita <Tomoya.Fujita@sony.com>
clalancette
deleted the
fujitatomoya/deprecate-static-single-thread-executor
branch
we should also deprecate
StaticSingleThreadedExecutorfrom the documentation.aligns with ros2/rclcpp#2598
Important
Do not backport this, because
StaticSingleThreadedExecutoris deprecated only inrolling.