They are equivalent. That doesn't mean that all tools will use XML semantics to test for equivalence.

A testing tool might require that the output is byte-for-byte equivalent to a known good output. Python's pseudorandomly determined hash function won't preserve that order across multiple runs.