Conversation
|
I am not sure I really understand the reasons behind this change. Why do we need to be able to use the standard Python logging library instead of From the PR description it seems you want to remove the ROS2 dependency, but this only removes the runtime dependency (this is still a ROS2 package that depends on the ROS2 build system). It is also still the preferred option. I am also not sure if this solves #8, as this is more of a toggle/fallback than really "injecting an logger". If you want to inject a arbitrary logger I would expect a common interface definition. The changes itself look good to me. |
The changes here are in the sprit of what I was thinking in #8. After this there is no actually need for ROS2 at all. Sure things are easier if you use |
e2f1628 to
6bf1153
Compare
|
From our conversation: this MR is part of a bigger move toward making this a pure Python package as there are no real dependencies on ROS 2 (also see #16) other than it using the |
This PR adds the ability to use python logging instead of rclpy.logging and removes the only real ROS2 dependency.
Fixes #8
Motivation and Context
The only real ROS2 dependency of this parser was the
rclpy.loggingmodule, so to remove that dependency but still make it possible to userclpy.loggingI added a selector for logging module dependent on if therclpy.loggingmodule was imported.Changes
Type of changes
Checklist
Testing
Run the parser on some example trees. If there are no exceptions thrown and the tree is built then it is working.