On Tue, May 25, 2010 at 2:21 PM, Patrick Bouffard
<bouffard@eecs.berkeley.edu> wrote:
> Tully,
>
> Actually what I'm doing right now is similar to the second scenario
> you describe, but I broadcast the transform twice, once with a
> ros::Time::now() timestamp, and immediately following I broadcast with
> a ros::Time::now() + ros::Duration(600) timestamp. Doing that, my
> frame appears in the right spot in rviz for the duration of the
> session (which is a playback of recorded data, that does not include
> the transform I'm talking about). So what you're saying is that what's
> happening is that rviz is seeing the present-dated transform, and
> using that for the duration of the session, while my second
> future-dated transform is actually not being used. The problem is, if
> I comment out the second, future-dated transform broadcast, I get the
> "No transform from [/grey/camera] to frame [/enu]" message again.
> Could it be that rviz actually is using both the present and the
> future dated transforms and 'interpolating' for the time in between?

What time base are you using?

Do you specify the rosbag play --clock option?
-- 
 joq