Hello ROS (1) developers. Recently, I was facing some issues due to the fact that I had more than one /tf_static publisher in my system. I think that’s not completely unusual, since I’ve seen several people trying to do the same.
The problem is the following: the /tf_static topic should have all publishers latched, which is good for latecomers online, but not sufficient when replaying a bag - then you will only get the transforms from the last publisher that published to /tf_static. What I (and more people) would expect is to get all the static transforms. That is not accomplishable through latching.
I tackled the problem by creating the static_transform_mux node that sits there all the time, remembers everything it saw on /tf_static, and tries to always be the last publisher on the topic publishing all the transforms it saw.
What do you think about this approach? Do you think it’s worth trying to publish it as an official ROS package? Or should I instead send a PR to tf2_ros or rosbag?
Thank you very much for your opinions!