Comment 19 for bug 390563

Just to follow up on Robert's "further investigation". iter_interesting_nodes filters interesting keys, so the line:
        for record, items in chk_map.iter_interesting_nodes(chk_bytes,
                        self._chk_id_roots, uninteresting_root_keys):
                for name, bytes in items:

Those items have already gone through a set difference.

Still need to probe into why it seems we are getting more than we should. *my* contention is that something about the *values* is genuinely different because of the buggy parents issues. (one side claims more per-file parents than the other side does.)