Suppose you are answering a code-golf challenge in a language whose interpreter can only use UTF-8. Say you have a program that uses some code points in the range 128-255, but not higher. Would it then be permissible to encode your answer in Latin-1 or some other single-byte encoding?
I would argue that only encodings directly usable with the language interpreter should be allowed. Otherwise, one could compress any answer and then claim that it uses a zip encoding.
This question is motivated mainly by answers that use an APL code page. I suspect that not all modern APL interpreters are capable of executing source that uses this encoding (but maybe I'm wrong).