Michael Carroll

> If we use gz-msgs as components, wouldn't that mean the generated C++ code is now part of our public API? If the goal is to move away from shipping...

> Maybe this can be worked around by enforcing a protoc + gz-proto version combination? I think this is hard without vendoring `protobuf`. I think that protobuf, like many other...

Agree, I think that this is an interesting and useful feature. I agree that the name should also be changed. "rolling_buffer" seems like a suitable name. I believe that the...

We have found that the Python Debug story on Windows is a bit difficult. As far as I know, choco provides no mechanism for installing the debug python interpreter, which...

I'm pretty sure CI is consistently broken on the `ros2` branch.

> Something we've talked about along these lines would be to make the metapackages (ign-fortress, ign-garden...) export the IGN__VER CMake variables. That could then be used by ROS. But it...

This is going to be retargeted to galactic, and I have also made an effort to robustify tests here: https://github.com/gazebosim/ros_gz/pull/272

> CI is green, is this ready for review ? Not quite, I'm about half way there. It's a pretty massive forward port. Will remove draft status when it is...

Okay, this is now ready for a review. It's a pretty massive PR... I'm sorry, but it will bring ros2 completely up to date afaict.

> DCO needs fixing :O DCO is always a bit finicky around merge commits, I'll see what I can do.