How to simplify massive numbers of log files

edit

it is also highly likely that the log file is just empty...

Yes, I expect this as well.

@Cerin: debugging a crashing node is infinitely easier with gdb than looking at log files. See Roslaunch Nodes in GDB.

I've manually run rostopic and rosservice a few times, but not the thousands for all the UUID log files I'm seeing. If a named node listens of publishes to a topic, or calls a service, should that create one of these UUID log files too?

A new logfile is created for every new node that you start. The uuid is bound to the lifetime of a rosmaster.

To be clear: the nrs in the logs in your screenshot (the rosservice_...log ones) are unix timestamps, not uuids.

So how do I stop these from being created in separate files, or at least get them named to reflect the node/rostopic/rosservice calls that created them. As is, these log files are completely useless.

They are not completely useless, and I would appreciate it if we can keep this constructive.

The filename is generated here. There is no automated way right now, afaik, to ..

.. configure 'something' to generate filenames like you suggest (including command line args).

It could be done though, as any node that is passed the __log remapping argument will use whatever value is associated with it for a filename.

Also: how would you like the filename to be set for services/topics that are more complex than a single line of args (which would be just about anything that is not a toy example)? We could serialise the entire yaml contents of the request/msg, but that will lead to very long filenames.

The filename is generated here

for rospy, that is.

If using roslaunch or rosrun, the __log remapping is always supplied, otherwise the code generates a filename.