Olivier Michel
                                            Olivier Michel
                                        
                                    The advantages of using the C API instead of the C++ API are: - It saves the controller.py library from dependencies on libCppController and standard C++ libraries (only libController is...
Sure, we can fully drop Python 2 support.
I believe this new python API should be implemented from scratch as a new python module. Basically, it should simply load the libController shared library and call it's functions wherever...
I believe developing a python module that loads a C shared library and calls its functions is something pretty standard. Yes, you need to load libController which is named with...
Sorry, I meant `controller.dll` indeed. I will give the C implementation a go next week and keep you posted.
I believe it is an opaque C pointer, but can be casted to an integer anyhow.
I believe the `scale` parameter should be allowed only in the `Transform` node and in no other node, as it will cause various problems. Even in a simple `Solid` node,...
An idea to make things clean would be to introduce a `Pose` node that would be basically like a `Transform`, but without the `scale` field. The `Solid` nodes would then...
That's probably an issue caused by SWIG, yet another good reason to create a Python API which doesn't rely on SWIG, see also #2385 about it.
Since [macOS is given tier 3 support for ROS 2](https://discourse.ros.org/t/macos-support-in-ros-2-galactic-and-beyond/17891), I believe we shouldn't make too much efforts to fix this kind of bugs as we may end up bumping...