Receive-waiting methods like spin_once and spin_some in ROS

[akihiro-komorilab]

Hi,

In a single-threaded program, I would to subscribe to a message and do some other processing, like spin_once() in rclpy or spin_some() in rclcpp. How should I implement this in zenoh-python?

Thanks,

[Mallets]

zenoh-python is based on the Rust implementation which is already multi-threaded and asynchronous. If you want to be single-threaded, e.g. on embedded targets, you could use zenoh-pico (but it's in C).

[p-avital]

Hi @akihiro-komorilab,

The issue here is due to Python's garbage collection methodology. Notice that the example code doesn't follow best practice in that it runs its code in global scope. The issue with that is that sub and session get declared in that global scope, which only gets cleaned up when all children threads have been joined.

However, this joining can only happen upon the session's destruction (note that subscribers hold a strong reference to the session, making their destruction necessary to fully closing a session), when the threads spawned by the underlying Rust library will be cleaned up.

This problem doesn't appear as soon as these session and sub variables are declared inside a scope (for example, a main function which gets called after checking __name__ == "__main__" as is best practice), since leaving that scope will trigger the automatic cleanup.

We have a few higher priorities at the moment, but we'll eventually switch the examples to use a main function, as well as document this a bit better.

Thanks for spotting this :)

[akihiro-komorilab]

Hi,

Thank you so much for the details. I appreciate your help. I understand.

I was able to get the information I needed to do what I wanted to do. I was also able to incorporate the zenoh-python module into another application, which is what I originally wanted to do.
I am currently compiling development information into a web article and will let you know when it is published. (in Japanese though ;)

Thanks again!

[p-avital]

良いな、ぜひリンクを教えてください!　（全然漢字を読めないんだけど...）

[akihiro-komorilab]

Sure! I will tell you tomorrow. (Your use of Kanji is perfect already!)

[akihiro-komorilab]

Hi @p-avital
Here you are.
https://qiita.com/comocc/items/448deee5a0bf82b586ec
I had to mention you guys in the acknowledgements. :)
Thanks!