Dejan, These are definitely two of the larger problems with rosbag play at the moment and they keep getting pushed on account of being non-trivial to solve. With respect to a), as you mention, the time- and state-discontinuities make this reasonably complicated. Could you provide me with some more information on your particular use-case? The more I can understand about why it is that people keep requesting this feature, the better job of addressing it I can do. At the moment, I believe your best bet is going to be to have rosbag play publish the clock for you (using the --clock optioin), and then between each iteration of the loop, you are going to need to publish an empty message to the "/reset_time" topic. There is an open ticket (https://code.ros.org/trac/ros/ticket/1580) to have rosbag play do this time-reset for you. Unfortunately, a search for "reset_time" on the wiki does not provide any detailed information on what the ramifications of this time-resetting is going to be on your nodes. As long as your nodes are not too stateful, I think this will do the trick though. As for b) there is a plan to figure out the appropriate way of logging parameters (ticket: https://code.ros.org/trac/ros/ticket/2425). This will most likely be done by subscribing to a specified parameter list, and on appropriate invocation of rosbag play, re-setting these parameters over the course of replaying. In general you don't want to play back all of the parameters since any node running on your played-back data would end up fighting with the player over setting the parameter. However, until this support is added, your only option is to do a "rosparam dump" before you start recording and a "rosparam load" before you start playing back. This will only work if you are recording from nodes that do not do any parameter manipulation, but that is usually a true statement if you just record from sensors and you do not reconfigure them over the course of your recording. --Jeremy On Thu, Mar 18, 2010 at 5:13 AM, Dejan Pangercic wrote: > Dear rosistas, > I got following 2 questions: > > a)What is the recommendable way of playing rosbag files in a loop? To > my understanding a simple bash loop  will have issues with the time, > correct? > > b)Is there a plan to incorporate recording of ros parameters and ros > services? The former would be the only way to log the robot state > model for isntance. > > cheers, D. > -- > MSc. Dejan Pangercic > PhD Student/Researcher > Computer Science IX > Technische Universität München > Telephone: +49 (89) 289-17780 > E-Mail: dejan.pangercic@in.tum.de > WWW: http://ias.cs.tum.edu/people/pangercic > _______________________________________________ > ros-users mailing list > ros-users@code.ros.org > https://code.ros.org/mailman/listinfo/ros-users >