Hi Mark,
My point is that we can specify as part of the registration or a message send that an app (via the API) include a tag such as the pid of the sending process. The pid is tagged to particular instance of an app and the Hub assigns this ID to the recipients of the message so they all see the true originator of the message, but we don't need any ID information at the user level in the API, this is all in the Hub implementation.
At the moment we've got this conversation going to establish public and private IDs in the API and stuff for an app to worry about handling. A pid can certainly be hacked, but since the idea here is 'simple' I think we're just adding complexity for no real demonstrated threat. If the mythical "rm *" must be dealt with NOW, then so should some of the other radical use cases. I thought this all fell into the "no guaranteed delivery, no queueing, KISS, etc" principles agreed on long ago.
-Mike Received on 2007-04-27Z11:54:07