[ros-users] Unifying the structure of ROS

Brian Gerkey gerkey at willowgarage.com
Mon Jun 14 17:22:11 UTC 2010


hi Nick,

You definitely shouldn't nest packages; the ros tools assume that a
package does not contain a package.

The level of package granularity is up to you.  As a general
principle, we try to keep packages fairly small, because it's at the
package level where build dependencies are determined.  The result of
packages being small is, of course, that the number of packages is
large.  If you prefer a small number of packages, then the packages
will need to be bigger, at the cost of occasionally introducing
unnecessary dependencies.

	brian.

On Mon, Jun 14, 2010 at 8:06 AM, Dominick Vanthienen
<dominick.vanthienen at mech.kuleuven.be> wrote:
>
>
>
> Dear fellow ROS-users,
>
> I would like to make a package with some functional C class and examples of
> components/nodes that use that functionality.
> An example can be found on:
> http://svn.mech.kuleuven.be/repos/orocos/trunk/kul-ros-pkg/SSC32/
> The files in the src/ directory contain the functionality and
> orocosSSC32Example/ directory contains an example Orocos component which
> uses this functionality.
> In this example the second directory is also a package (nested packages)=>
> you'll have to add it to your $ROS_PACKAGE_PATH seperatly in order to get
> this working.
> Alternative could be:
> *a seperate package next to the package with the functionality, but this
> will lead to a huge number of packages: (one for the C functionallity, one
> for a rosnode example, one for an orocos component example...)
> *everything in one package: but then people need to install Orocos (RTT,
> because it will be in the manifest) for functionality that's not Orocos
> specific (then it has no use of making first a C class...)
> *...
>
> What solution would be most appropriate?
>
> Thanx for your comments
>
> Nick
>
>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
>



More information about the ros-users mailing list