Emacs and clang?

If I understand correctly the complexity of parsing the source and returning some sort or normalized representation of the code is going to be the same for a compiler as for a full-featured editor, so by definition the "simple" parser used by the editor, to be complete, will have to be the same complexity as the parser used by the compiler, so why not ditch the duplicate implementation (which won't be bug-for-bug compatible anyway) and use the compiler as the parser.