On Mon, Dec 5, 2011 at 6:10 PM, Trevor Jay <tjay@cs.brown.edu> wrote:

Actually, I think that answers my question. :) I was dense and didn't
quite grok on first glance that this was the beginning of being able
to include much less of ROS when making use of the message system.
This is a really really good direction. This can't come to my
distribution soon enough.

Glad to hear it. As you can see, my working usecase was "multicast ROS messages without [roscpp | rosmaster | ROS_* in environment | xml of any kind | rosmake]". Those dependencies have been substituted for boost::asio, a single ros library, and an assortment of fairly advanced CMake infrastructure known as "catkin". I don't expect this to work for everybody. I'd like to see more prototypes, and hear from those familiar with difficult integration scenarios. Where does this fall short? Don't be shy. Ask questions.

Also coming up, very soon, are boost::python bindings to ROS messages. This will enable yet more interesting patterns.

-t