True — but Unicode strings that are not bit-equal aren't collation-equal either, so that doesn't usually matter. A non-canonicalized Unicode string won't "find"

True — but Unicode strings that are not bit-equal aren't collation-equal either, so that doesn't usually matter. A non-canonicalized Unicode string won't "find" its canonical counterpart in a hashtable. IEEE754 floats (and doubles) are kind of unique among scalar types, in that the answer to "are these equal" in pretty much every programming language is delegated directly to the CPU to answer; and, in obeying IEEE754 semantics, the CPU's definition of equality for FP numbers makes things equal that aren't bit-representation-equal.