[tremon]

For completeness sake: the confusing part is that there is (used to be) a constant-length encoding that uses the exact same codepoints as UTF-16, but doesn't allow the variable-length extensions. That encoding is called UCS-2 and although deprecated, is the reason why many people think UTF-16 is constant-length.

[cryptonector]

Not the same codepoints as UTF-16, but a subset. But yes, UCS-2 is the source of all this evil.