If we are going to standardize this we should be using
existing standard
color spaces that are widely implemented.
I had to go re-read the XEP to see what it is you mean here. I do consider
it odd that the spec mentions HSLuv in normative text. Since the algorithm
only generates a hue angle, it is a sensible input to any HSL
implementation, not just HSLuv. Since the XEP leaves S and L to be
implementation-defined, the difference between HSLuv and any other HSL will
not result in non-compliance in any case.
In addition, as a small editorial note, section 9.4 still refers to CbCr
even though the XEP does not anymore.
Section 13.1 contains r,b,g output values, even though these cannot be
calculated without at least an S and L, which are not specified. Probably
only angle/hue should be listed since it is the actual output of the
algorithm.