Thanks - that's very interesting (@LAST_MATCH_START).
Sorry I wasn't clear about the use case, but I'm now curious about the representation - I'm aware there are many ways to test the regexp status.
I had been using $& to test for the general match state, but that can perform badly.