Feature description
At present, ROS 2 driven benchmarks use separate processes for playback, execution (via launch files), and recording. This is general and flexible, but incurs all the network traffic and OS overhead of a multi-process ROS 2 graph.
We could do better for composable nodes under test. We could have a fully-composed, single-process ROS 2 benchmark. Benchmarks could run faster.
Implementation considerations
Composable rosbag2 playback and recording nodes are a ROS 2 Jazzy feature.
Feature description
At present, ROS 2 driven benchmarks use separate processes for playback, execution (via launch files), and recording. This is general and flexible, but incurs all the network traffic and OS overhead of a multi-process ROS 2 graph.
We could do better for composable nodes under test. We could have a fully-composed, single-process ROS 2 benchmark. Benchmarks could run faster.
Implementation considerations
Composable
rosbag2playback and recording nodes are a ROS 2 Jazzy feature.