Hi Mike, If you continue to have a problem, please open a ticket with the roslaunch log from the three scenarios you describe. The log file will contain the exact, expanded settings roslaunch is using. - Ken On Mon, Sep 13, 2010 at 9:25 AM, Mike Purvis wrote: > Ok, thanks. I'll take another look. Just to clarify, though: > 1. Remote launch with the machine tag does work. > 2. Local launch without the machine tag does work. > 3. Local launch with the machine tag does not work. > It seems a bit bizarre that an issue with ~ paths would manifest itself as a > problem in the third case and not the first one. (The second case is > unaffected, since there is no machine tag.) > M. > > On 13 September 2010 11:24, Ken Conley wrote: >> >> Hi Mike, >> >> Given that this is a path issue (basically, ROS_ROOT isn't being set >> properly), David is probably on the right track with the ~ issue. If >> expanding it to the full path works, please let me know and I'll file >> a ticket. >> >>  - Ken >> >> On Mon, Sep 13, 2010 at 8:15 AM, David Dröschel >> wrote: >> > Hi Mike, >> > >> > Is it possible that the "~" is substituted on the local computer (by the >> > local user home)? I'm just guessing... >> > >> > I remember something like that when i was setting up our machines here >> > but i don't have access to them atm >> > >> > Perhaps you can give it a shot. >> > >> > David >> > >> > Mike Purvis wrote: >> >> Hi David, >> >> >> >> Yes, I believe it's all correct, in part because it totally works when >> >> /actually/ run remotely. But for reference, here's the relevant line >> >> of machines.launch: >> >> >> >> >> >> > >> ros-root="~/ros/ros" >> >> ros-package-path="~/ros/stacks:~/ros-clearpath:~/umd-ros-pkg" >> >> default="never" timeout="5.0" /> >> >> >> >> >> >> Thanks! >> >> >> >> >> >> >> >> On 13 September 2010 10:31, David Dröschel > >> > wrote: >> >> >> >>     Hi Mike, >> >> >> >>     have you set the ROS_PACKAGE_PATH with the ros-package-path >> >>     attribute in >> >>     the machine tag of cpr-demo-01? >> >> >> >>     You have to do that if the ROS_PACKAGE_PATH of cpr-demo-01 is >> >> unequal >> >>     the one from your desktop. >> >> >> >>     David >> >> >> >> >> >> >> >> >> >>     Mike Purvis wrote: >> >>     > Hey, >> >>     > >> >>     > Up until now, we've been doing an awkward scheme of one >> >>     roslaunch file >> >>     > per machine, which has meant that they only ever have to launch >> >>     local >> >>     > nodes. However, I'm trying to clean it up to take advantage of >> >>     > roslaunch's remote-start abilities. >> >>     > >> >>     > I have it mostly working, but there's one little annoyance: >> >>     > >> >>     > >> >>     >   >> >>     > >> >>     >   >> >>     >     > >> output="screen"> >> >>     >       >> >>     >     >> >>     > >> >>     >     > >> name="joy_horizon" >> >>     > output="screen"> >> >>     >       >> >>     >       >> >>     >     >> >>     > >> >>     >     > >> name="horizon" >> >>     > machine="cpr-demo-01" respawn="true" output="screen"> >> >>     >       >> >>     >       >> >>     > >> >>     >       >> >>     >       >> >>     >       >> >>     >       >> >>     >     >> >>     >   >> >>     > >> >>     > >> >>     > Now, this works great when I launch it on my developer desktop--- >> >> I >> >>     > get a joy_node and a joy_horizon coming up locally, and horizon >> >>     comes >> >>     > up on the cpr-demo-01 machine, and I can joystick around our >> >>     chassis. >> >>     > >> >>     > However, I like to be able to plug the joystick into the >> >> cpr-demo-01 >> >>     > machine and actually run it all from there, too. It seems like >> >> this >> >>     > should be possible, but when I try it, the horizon comes up with >> >> a >> >>     > Python include error--- it can't find roslib, of all things. This >> >>     > error doesn't show when I run it remotely, and it's also fine if >> >> I >> >>     > remove the machine="cpr-demo-01" part. >> >>     > >> >>     > Is this documented behaviour, or am I doing something wrong? It >> >>     would >> >>     > be really convenient if this could be made to work without having >> >> to >> >>     > maintain separate files for whether the joystick is onboard the >> >>     robot >> >>     > or off-board. >> >>     > >> >>     > Thanks, >> >>     > >> >>     > Mike >> >>     > >> >> >> >> ------------------------------------------------------------------------ >> >>     > >> >>     > _______________________________________________ >> >>     > ros-users mailing list >> >>     > ros-users@code.ros.org >> >>     > https://code.ros.org/mailman/listinfo/ros-users >> >>     > >> >> >> >> >> >>     _______________________________________________ >> >>     ros-users mailing list >> >>     ros-users@code.ros.org >> >>     https://code.ros.org/mailman/listinfo/ros-users >> >> >> >> >> >> >> >> ------------------------------------------------------------------------ >> >> >> >> _______________________________________________ >> >> ros-users mailing list >> >> ros-users@code.ros.org >> >> https://code.ros.org/mailman/listinfo/ros-users >> >> >> > >> > >> > -- >> > David Droeschel, M.Sc. >> > >> > Autonomous Intelligent Systems Group >> > Computer Science Institute VI >> > University of Bonn >> > Römerstr. 164 >> > 53117 Bonn, Germany >> > Phone: +49 (0)228 73-4434 >> > Fax:   +49 (0)228 73-4425 >> > >> > >> > _______________________________________________ >> > ros-users mailing list >> > ros-users@code.ros.org >> > https://code.ros.org/mailman/listinfo/ros-users >> > >> _______________________________________________ >> ros-users mailing list >> ros-users@code.ros.org >> https://code.ros.org/mailman/listinfo/ros-users > > > _______________________________________________ > ros-users mailing list > ros-users@code.ros.org > https://code.ros.org/mailman/listinfo/ros-users > >