Code review comment for lp:~nick-dedekind/mir/trusted_sessions

Looking at the client API:

    MirTrustSession* mir_connection_create_trust_session(MirConnection* connection);

I would expect "create_trust_session" to take parameters (such as the event callback) and create the server-side trust session (which would implicitly make it an async call).

AIUI this just creates a client-side object that still needs to be initialized before the server-side trust session is created using "mir_trust_session_start".

Is there any reason to separate out these create/set_event_callback/start calls?

~~~~

mir_trust_session_add_trusted_session

This is an async call - so presumably it is expected to be invoked after the trust session is started?

Does that imply a round trip for each pid added to the trust session? If so would it be better to accept a list of pids?

Can it be called on a created (but not started) trust session?

~~~~

Is there a missing mir_trust_session_remove_trusted_session() for revoking trust?

~~~~

mir_trust_session_set_event_callback

The only state changes appear to be "start" and "stop" - which are under the control of the client. So does this do any more than duplicate the functionality of the WaitHandles returned by the start/stop functions?