On 10/02/2010 12:55 PM, Ivan Dryanovski wrote:
>
> Another thing to consider is a convention about out-of-range readings.
> If I recall correctly, LaserScan drivers currently return "0" for
> out-of-range.
>
> Ivan Dryanovski

Ivan, we've actually changed this a little bit for the upcoming distribution (D). Though not fully implemented and 
reviewed yet, we've agreed on using NaN and Inf values for invalid and out of range measurements. Stereo point clouds 
already use NaNs to express invalid disparities.

We'll try to get this through a formal API review process soon. Obviously, we'll send out invitations to other 
developers/users to join the talks.


Cheers,
Radu.