Fix the unicode test string for opensplice rmw implementation

[BMarchi]

As the title says, this fixes a particular test for opensplice.
I don't think this is a proper solution for the issue, but it keeps CI happy. What I was able to understand and find, is that opensplice caches the topics that are created. I know we are using a deb, but I took a peek to their community edition and it seems that the errors matched just fine. You can see that error here.
As I was saying, when one creates a publisher/subscriber, it reaches a point where it creates a topic, calling this function

https://github.com/ADLINK-IST/opensplice/blob/c98e118a2e4366d2a5a6af8cbcecdf112bf9e4ab/src/kernel/code/v_topicImpl.c#L477

Specifically, this is part of our interest

https://github.com/ADLINK-IST/opensplice/blob/c98e118a2e4366d2a5a6af8cbcecdf112bf9e4ab/src/kernel/code/v_topicImpl.c#L527

Here, it makes use of a lookuptable for topics that lives within the kernel.

https://github.com/ADLINK-IST/opensplice/blob/c98e118a2e4366d2a5a6af8cbcecdf112bf9e4ab/src/kernel/code/v_topicImpl.c#L532

That function is the one that checks the types of the new topic and the one it found (since the names are the same).
Digging a little bit in the rmw implementation for the publisher, I saw that we do call a delete_topic function for the participant

https://github.com/ros2/rmw_opensplice/blob/f9997162920857a46a782ad1fbf7b876f6ba05bd/rmw_opensplice_cpp/src/rmw_publisher.cpp#L274

And it seems that there's no issue in "deleting" the topic, otherwise an error/exception should show.
When a node is destroyed, in rmw_destroy_node, I found a curious comment.

https://github.com/ros2/rmw_opensplice/blob/f9997162920857a46a782ad1fbf7b876f6ba05bd/rmw_opensplice_cpp/src/rmw_node.cpp#L381-L386

The interesting thing is that it seems that we are somewhat aware that some stuff could be in memory yet even if they shouldn't.
Does anyone have a clue if this is a known issue or we are missing a function call in the destroy_publisher function?

• Opensplice jobs:
• Linux
• Arch
• macOS
• Windows

[mjcarroll]

Thanks for debugging this one @BMarchi

[BMarchi]

Should we call someone from opensplice to give us an insight of what's going on?

[hidmic]

Should we call someone from opensplice to give us an insight of what's going on?

I think we should, but that fix won't land before Eloquent is out so let's merge this one!