On Wed, Jun 22, 2011 at 9:48 AM, Nicholas Butko <nbutko@ucsd.edu> wrote:
> Damon,
> One option is to take the course of "When in doubt, follow the behavior of
> roscpp." In that case, when a second node registers with the same name, the
> first node dies. For example, if you run
> {{{
> rosrun roscpp_tutorials talker
> }}}
> twice, the second instance behaves normally, but the first instance shows
> {{{
> [ INFO] [1308753854.115622000]: hello world 72
> [ INFO] [1308753854.215643000]: hello world 73
> [ WARN] [1308753854.285720000]: Shutdown request received.
> [ WARN] [1308753854.285791000]: Reason given for shutdown: [new node
> registered with same name]
> [ INFO] [1308753854.315770000]: hello world 74
> }}}
> Just a thought. What do you think?

That seems like the right behavior to me. It allows a node to be
restarted successfully even when the previous instance has not
completely shut down. That can be useful for failure recovery.
-- 
 joq