Pubsub nodeid question

This is revisiting a question I started on a previous thread which I marked as answered.

After some thought and looking again at XEP-0060 I wanted to reopen the part of my question about inconsistent use of nodenames in pubsub commands.


actually creates a node called “/test” (I’'m happy so far but:

could never work in wildfire’'s current pubsub implementation.

  1. It assumes that the client that created a node is the same as the client that wants to perform other commands on it. If the client that created the node is the only thing that knows the node name of the item it created, in general, this makes it much more difficult for other clients to discover the real node name.

  2. In my use-case, I am creating nodes based on the current user’'s jid, so, if for example another user wants to subscribe or get items for another person, his client needs to make an assumption about the transformation from requestedNodeID to actualNodeID. Admittedly, this may be a simple as making sure the node name has a leading slash, but seems to me this a dangerous assumption in general, because different implementations of pubsub may escape the requestedNodeID in arbitrary ways. Or did I make that up?

So… to cut a long story short, I would request that all pubsub commands can accept what I describe as ‘‘requestedNodeID’’, if for no other reason than this functionality seems to be the assumption in the examples of XEP-0060

Am I making sense?

Chris Denham