Let's try to revive this discussion... +1 from me on the proposed spec. The transition plan sounds adequate to me, perhaps even overkill. But I'm not a user of LaserScan or Range, just PointCloud2, which already uses NaN and Inf. Cheers, Patrick On Fri, Oct 14, 2011 at 12:53 PM, Ken Conley wrote: > This has been assigned number 117 and is viewable here: > > http://www.ros.org/reps/rep-0117.html > > cheers, > Ken > > On Thu, Oct 13, 2011 at 8:28 PM, Chad Rockey wrote: > >> There is a need to communicate special values within the distance >> measurement sensor messages (Laserscan, Range, PointClouds). In this REP, >> I propose using the standard floating point values of -Inf, NaN, and +Inf >> to represent more information within the current messages. Previous discussion is >> available on our wiki page ( >> http://www.ros.org/wiki/fuerte/Planning/Drivers) and in ticket #4367 ( >> https://code.ros.org/trac/ros-pkg/ticket/4367). >> >> Attached is my first draft of this REP. The Drivers SIG has reviewed >> this REP, and we now seek community input. Our main concern with this REP >> is transition strategy. We wish to provide minimal disruption with this >> change (no message migration, minimal client code changes); however, in >> certain cases NaNs may propagate through systems. Our primary concern >> is how to adequately prepare the community for this change. >> >> To mitigate transition problems, we propose to add a node, nodelet, and >> library that will clean the target messages of NaNs and Infinities and >> revert the new measurements to the legacy 'max_range + 1' behavior. This >> tool would only be supported for one release and removed for the Groovy >> release. >> >> Thank you for your feedback. >> >> - Chad >> >> _______________________________________________ >> ros-developers mailing list >> ros-developers@code.ros.org >> https://code.ros.org/mailman/listinfo/ros-developers >> >> _______________________________________________ >> 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 > >