The idea is to make the replication protocol a bit more flexible, in a way similar to what 5c837dd has done, but for repl_scanner.l. This will also allow to avoid any problems like what has been fixed in aa41bc7 where the SHOW commands used in libpq have to be capitalized. Personally, I have pested about the lack of flexibility a couple of times when running tests using psql..
I am parking that in the next CF.
Given that the protocol really isn't intended for "manual consumption", do we really want that? Not that it adds a lot of complexity, but still. It certainly makes the code completely unreadable. And since any program using it should figure out pretty quickly that it's not working if they us the wrong casing...
