>Pablo, Is there any sane possibility that the compiler could, at least sometimes, raise an IncompleteSyntax subclass of SyntaxError when code is merely incomplete, as opposed to wrong?

I'm assuming you mean the parser (when the compiler runs the code must have been correct or otherwise the parser would have raised before).

In case of the parser, I'm not sure but the answer is generally no. The parser doesn't have a concept of "partial Syntax" so that should be defined. There is also the problem that some of these syntax error come from the tokenizer which has its own rules. We could probably do more in the tokenizer than in the parser but this would add considerable complexity in the general case.