Michel Hidalgo

Long term, I agree with @ivanpauno. To have RMW events being waited on a background thread and dispatched to an `asyncio` loop is probably the simplest approach. Incoming messages from...

> I honestly don't know how to change this since the templating seems quite complex. Some advice would be appreciated. @Flova you could check for instance type right before [this...

The first error is expected, `launch_test` is ROS agnostic and will not automatically prepend the default launch description that's required for many `launch_ros` actions to work. One may argue that...

@Parallel-Hao I'm curious, how did you manage to use Foxy binaries? AFAIK there aren't any yet, we haven't even started with preliminary releases. How did you install ROS 2?

@Parallel-Hao I'm sorry for the delay. The problem seems to be `python` itself when run from a non-interactive shell (see [here](https://stackoverflow.com/questions/974189/how-to-send-a-sigint-to-python-from-a-bash-script)). Reinstalling `signal.default_int_handler` does the trick. It's unclear to me...

> How does that SO link explain this? We register the signal handler in C. They just say you need a python signal handler in the SO thread. The relevant...

We'll need to sync up `rclcpp` and `rclpy` in this regard then. I'll try to put together a patch, but I don't think it'll make it to Foxy.

Alright, IMHO this should get done for Galactic, but I can't say I'll have time to rework `rclpy`'s signal handling to match that of `rclcpp` in the near future. FYI...

> Adding a default signal handler to rclpy that raises KeyboardException sounds trivial, so I will mark this as help wanted (feel free to undo that). Hmm, IIUC the conclusion...

> why not just making sure KeyboardInterrupt is raised in the signal handler we're installing? Having a library silently changing default (and expected) behavior isn't a good idea. > The...