[geometry] Use of SameValueZero in "validate and fixup" doesn't make sense

[bzbarsky]

https://drafts.fxtf.org/geometry/#matrix-validate-and-fixup says things like:

a and m11 are both present and SameValueZero(a, m11) is false.

but a and m11 are both unrestricted double IDL members, so this is passing two doubles to a function that expects two ES values. It should either pass two Number values constructed from a and m11 or it should be using something defined to work on doubles.

[dirkschulze]

@bzbarsky What is the preferred way? Constructing ES number value from a unrestricted double or does WebIDL or another spec provide means to compare unrestricted doubles for "equality"?

[bzbarsky]

I have no idea, actually. Constructing ES number values is probably simple enough...

[tabatkins]

The ES spec has to deal with this sort of thing as well, because it draws a similar distinction between abstract numbers and ES number objects. @littledan, @bterlson, any advice?

(Tagging them in because I think I recall them being involved with some recent edits to make the ES spec more specific in this regard.)