Serialize encapsulation header on buffer aware topics - #891
Conversation
Signed-off-by: Miguel Company <miguelcompany@eprosima.com>
Signed-off-by: Miguel Company <miguelcompany@eprosima.com>
wjwwood
left a comment
wjwwood
left a comment
There was a problem hiding this comment.
Makes sense to me. I also think a backport is appropriate. Might need to run it by the PMC meeting and probably post about it to discourse.
|
Pulls: #891
|
|
Tick the box to add this pull request to the merge queue (same as
|
|
@mjcarroll @sloretz As discussed in the PMC meeting, I added a log error for the version mismatch when reading the encapsulation header |
Signed-off-by: Miguel Company <miguelcompany@eprosima.com>
375754d to
509fb5f
Compare
|
New CI:
|
Signed-off-by: Miguel Company <miguelcompany@eprosima.com>
|
New CI with just
|
|
@mjcarroll I think we can merge this into rolling |
|
@Mergifyio backport lyrical |
✅ Backports have been createdDetails
|
4 participants
Description
While creating a native DDS consumer for rosidl buffer custom topics, I discovered that the encapsulation header required by the RTPS standard is not added for the reader-GUID-based topics.
This PR adds the encapsulation header when publishing, and consumes it when taking.
Is this user-facing behavior change?
No, but the custom serialization changes.
I would nonetheless consider backporting to lyrical.
Did you use Generative AI?
No.
Additional Information
I verified that the GPU buffer transport tutorial works with these changes