Revision history [back]

Another potential solution is to write the second part of your system (the part that takes input from the keyboard) as a separate ROS node that communicates with the original node. Then, you don't have to drop out of the spin(). This doesn't solve the problem of knowing when rosbag is done, though.

One hack that might work is to record a second bag file done.bag with a single message in it on a new topic (let's call it done). Then play the bags back-to-back: rosbag play subset.bag done.bag. When you see the message on the done topic, then you know that the first bag is finished, and that you can start listening for command-line input (using the method in the first part of this answer).

A bit kludgey, but it might work.