Code review comment for lp:~robertcarr/mir/implement-client-credentials

This isn't quite what I expected.

Walking through app launch process: shell launches an app, and knows the app's PID. When app connects to Mir, this authorizer will check the app's PID with shell. Shell knows the info of the application with that PID, so can decide to accept or reject.

However after that, how can shell make a connection between the generated Session and the application list it has internally? PID isn't attached to Session, Session only has a "name" property to identify it uniquely. But what sets "name"?

I had expected that shell would set the name that Mir attaches to the Session.