Compact output still has XML semantics
Empty-element spelling, declaration layout, and insignificant indentation may normalize. Character data, CDATA content, comments, processing instructions, and namespace identities remain part of the parsed document.