Let's start from the beginning:
In ROS 1 noetic we received a lot of warnings if the AauSpatialReasoner published TF information with the same timestamp.
To tackle that we introduced a _last_tf_timestamp member:
https://github.com/RVMI/skiros2/blob/ed5bc907f8509f8223bf5c75ff2b7bcf9ff056e7/skiros2_common/skiros2_common/core/world_element.py#L68
Which is used by the reasoner:
now = self._node.get_clock().now()
# Prohibits publishing with the same timestamp - triggers warning in noetic
if e._last_tf_timestamp == now:
return
tf = e.getData(":TransformMsg")
tf.header.stamp = now.to_msg()
e._last_tf_timestamp = nowIn ROS 2 we obviously use the rclpy.Time object and there were pickle errors when deep-copying it. Deep-copying is for example used when doing spatial transformations with the reasoner since that directly changes the element.
I tried to reproduce this pickle error, but it worked then to do the deep-copy.
If nobody else can reproduce this, one can at least check if we need to save the last tf timestamp in ROS 2. If not, we can just remove it.
Let's start from the beginning:
In ROS 1 noetic we received a lot of warnings if the
AauSpatialReasonerpublished TF information with the same timestamp.To tackle that we introduced a
_last_tf_timestampmember:https://github.com/RVMI/skiros2/blob/ed5bc907f8509f8223bf5c75ff2b7bcf9ff056e7/skiros2_common/skiros2_common/core/world_element.py#L68
Which is used by the reasoner:
In ROS 2 we obviously use the
rclpy.Timeobject and there were pickle errors when deep-copying it. Deep-copying is for example used when doing spatial transformations with the reasoner since that directly changes the element.I tried to reproduce this pickle error, but it worked then to do the deep-copy.
If nobody else can reproduce this, one can at least check if we need to save the last tf timestamp in ROS 2. If not, we can just remove it.