Legacy URLs are preserved by attempting to parse the new URL format and, on failing that, it'll attempt the legacy one. Tests had to be updated such that the correct things are asserted against each other.