[Standards] PEP private nodes

Ralph Meijer jabber.org at ralphm.ik.nu
Mon Apr 9 12:48:21 CDT 2007


On Mon, 2007-04-09 at 11:13 -0600, Joe Hildebrand wrote:
> On Apr 6, 2007, at 1:20 PM, Tomasz Sterna wrote:
> 
> > And it doesn't go much along the lines of our "discovery over
> > convention" preference in design principals.
> 
> But it would work. :)
> 
> Also, it fits a bunch of the other design principals.  Simple.   
> Explicit.  Privacy matters.

I have been disconnected for a couple of days and have not read
everything on the standards list yet, but the following idea popped up
in my head, which seems to be similar to Joe's suggestion, but with a
slightly different angle.

When we introduced pubsub, we said that the 'node' namespace for pubsub
is the same as that for disco. I.e. if you have a pubsub node tied to an
entity (pubsub service, but also a user account in the PEP case), you
can disco it with the same node identifier.

Now, it seems that we need a replacement for iq:private and PEP has some
nice protocol we could reuse, but it doesn't really fit the PEP model.
So here my idea:

What if we just write a simple protocol for doing Private Storage. Just
an iq to write a snippet of XML to a named node. Also, a simple iq to
read the snippet back from the node. You have automagic constraints on
who can read, write. Very clear.

Now, imagine the namespace for these Private Storage nodes also
coincides with both disco and pubsub, then you can use the niceties of
pubsub to get your notifications using that protocol.

The point being that you can use stuff from different protocols to think
up a meaningful service.

Opinions?

-- 
Groetjes,

ralphm



More information about the Standards mailing list