Conversation
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
…shed Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
I just took a quick look, and it looks like there is already an (build) indirect dependency between
|
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
…ervice introspection Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
…tion configure Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
Signed-off-by: Brian Chen <brian.chen@openrobotics.org>
jacobperron
left a comment
There was a problem hiding this comment.
The approach looks good to me. Besides the string manipulation related to service typesupport, the code looks straight-forward 👍 I also agree with Ivan's comments.
I realize you haven't focused on error handling yet, but when you do you should look out for the following (which might help uncover bugs downstream):
- Always check the return values of functions (this isn't happening in many places)
- Make sure to do any necessary memory deallocation before returning from functions (pay attention to places where you are returning early due to an error).
- Check arguments to functions before using them (e.g. if they are null or invalid).
|
Closing in lieu of #997 to keep all service introspection work under a single branch name |
|
@ihasdapie I think you meant to close this PR (replaced with #997), so I'm closing it now. |
|
@ivanpauno Yup, thanks! I had forgotten. |
This PR is a prototype for the service introspection REP (ros-infrastructure/rep#360) implementing the proposed serialized service event approach.
Few items of note
rclonrosidl_typesupport_introspection_candrosidl_typesupport_crcl/rcl/src/rcl/introspection.c
Lines 57 to 119 in 9c0b42e
client_idis not populated on client send request events due to thewriter_guidnot being accessible from theclient_send_requestscope. PRs will be made to thermwimplementations to make this accessible.For instructions on how to build and run this see the meta-ticket ros2/ros2#1285
Resolves ros2/ros2#1285