[ros-users] Header msg anomaly in unstable

[Ken Conley]

On Fri, Jan 28, 2011 at 2:26 AM, Ruben Smits
<ruben.smits at mech.kuleuven.be> wrote:
> Hi,
>
> I just noticed that the roslib/Header.h got deprecated in favor of
> std_msgs/Header.h. Shouldn't this change also be resembled in the msg files?
>
> Header stamp --> std_msgs/Header stamp
>
> as it is for all other msg types. It would also make it very clear that if you
> want to use the Header msg type, you need to depend on the std_msgs package.
> (This hidden dependency actually made me run into the anomaly)

I'm not sure I fully understand.  Before there was a hidden dependency
on roslib.  Now there is a hidden dependency on std_msgs.  Both cases
seem equally bad.  rospy masks the hidden dependency but I believe
roscpp used to, then stopped, masking the roslib include dependency.

We do hope to vanquish Header completely, but the change you recommend
would change the md5sums of many messages, which would make
Diamondback/C Turtle wire incompatible.  We could, of course, change
the md5sum calculation, but that would perpetuate the special-ness of
Header in a different way.

FYI: there is additional discussion of this in REP 100:
http://www.ros.org/reps/rep-0100.html

 - Ken

> Ruben
>
>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>