The name of the PR suggests to me that we are changing the demos to inherit from node in this PR; for clarity when searching PRs in the future could you simplify it?

Sure, do you have a suggestion?

Sure, do you have a suggestion?

"Fix import statement and usage for rclpy.node.Node"?

This is really unfortunate because it adds another line to all of the examples unnecessarily (imo). It is also asymmetric with what we want to do with the C++ API, i.e. we want rclcpp::Node rather than rclcpp::node::Node. I don't understand fully why @dirk-thomas needed to do what he did in ros2/rclpy#147 (comment), but I think it's a step backwards for the usability of the rclpy API.

"Fix import statement and usage for rclpy.node.Node"?

Sounds good to me

This is really unfortunate because it adds another line to all of the examples unnecessarily (imo). It is also asymmetric with what we want to do with the C++ API, i.e. we want rclcpp::Node rather than rclcpp::node::Node. I don't understand fully why @dirk-thomas needed to do what he did in ros2/rclpy#147 (comment), but I think it's a step backwards for the usability of the rclpy API.

Yeah I agree, I'm proposing this fix mostly to get the demos back to a working state but don't think users should have to import this explicitly in the long-term.
I'd like to merge this before next week but can open a follow-up issue to revisit this before the release. How does that sound ?

+1 for getting things working again, but we should try to address this before the release I think.