Re: What is actually locked?

Perhaps more to the point, the issue is irrelevant because it
deals with a server-side implementation instruction rather than
interoperability between client and server.  The client should
be relying on the answers it gets back from the server, not on
an obscure requirement on how to implement locks within the server.

Regardless of how the server implements those locks, it will
respond with success or fail such that the client will be informed
whether or not it needed the lock token.  The client never needs
to know whether the URI corresponds to a binding, so adding
bind-specific requirements merely obfuscates the protocol.

....Roy