Big-endian vs. little-endian in the context of bit-level encoding

The intention of the messages I posted above was to show that the current implementations (PyUAVCAN, Libuavcan, Libcanard) are spec-compliant. I did not mean to imply that the currently specified behavior is optimal.

You are saying that the culprit is here:

Do we want to change it? The affected sections of the specification are:

  • 3.7.1.2 Bit and byte ordering
  • 3.7.3.1 General principles
  • 3.7.5 Composite types
  • 4.2.3 Examples