remove duplicate services and msgs

[v-lopez]

iterate_packages might return ROS2 paths for already found ROS1 pkgs

This results in duplicated entries in data and double definitions in the generated cpp code.

This happens to me building for instance this package: https://github.com/pal-robotics/pal_msgs/tree/ros2-tests/pal_common_msgs

I am trying to have both ROS1 and ROS2 msgs on the same branch, and need to do some package.xml and CMakeLists hacking, which might be tricking iterate_packages.

I have 3 workspaces:

ros1_ws extending /opt/ros/noetic, built with caktin_make install
ros2_ws extending /opt/ros/foxy, built with colcon build
bridge_ws, extending ros1_ws and ros2_ws install dirs.

And rosmsg.iterate_packages(rospack, rosmsg.MODE_MSG) returns:

 ('pal_common_msgs', '/root/msgs/ros1_ws/install/share/pal_common_msgs/msg'),
 ('pal_common_msgs',
  '/root/msgs/ros2_ws/install/pal_common_msgs/share/pal_common_msgs/msg'),

Am I doing something wrong? I was surprised that this did not happen with other messages that are both installed in ROS1 and ROS2.

[v-lopez]

I have a similar fix for actions for when #256 is merged, or if this is merged first I'll push there.

-        if match:
-            actions.append({
+        action_data = {
                 'ros1_name': pair[0].message_name,
                 'ros2_name': pair[1].message_name,
                 'ros1_package': pair[0].package_name,
                 'ros2_package': pair[1].package_name,
                 'fields': output
-            })
+            }
+        if match and action_data not in actions:
+            actions.append(action_data)

[sloretz]

Is it guaranteed that the first message found will be the ROS 1 version and not the ROS 2 version? If not then this check might not be enough.

[v-lopez]

I don't think it's guaranteed, so you're right.

I've reimplented this, please take a look.

[v-lopez]

Friendly ping, could we move this? Is it going to make it to foxy or just galactic an newer?

[Timple]

Probably not, galactic is EOL already...