[Standards-JIG] jep-0060 $home

bernhard bz at datenkueche.com
Tue May 30 16:25:02 UTC 2006


Hi
> How is ejabberd doing it? Would this help?
>   
In ejabberd the node attribute isn't an id.
It's more a method like xpath or xpointer or directory+filename to 
access a node.

Every user has write access to
"/home" +servername+"/"+jidprefix+"/"+whateveryouwant

Example: pubsub at jaim.at -> /home/jaim.at/pubsub/...

There is a node "home" with a childnode "jaim.at"
"jaim.at" has at the beginning no childnodes but the user "pubsub" has 
writeaccess to make a
childnode with the name "pubsub" to the node "jaim.at"...

If an instant node is generates it is generated as 
"/home/jaim.at/pubsub/234efdcsdf4r23"

As long as a node is not additional assigned to a collection or an other 
node the node attribute is unique and makes sense.
If nodes are moved, it isn't a good idea to take the node attribute as 
an id.
I did not implement this, but I doubt they use the node attribute as an id.
The "real" id is hidden.

In jep-0060 I couldn't find the information that the node attribute is 
an id.

For me the ejabberd method makes sense.
They didn't implement collections - but I still don't see the need for 
collections.
Collections are similar to directories in the filesystem.
If a pubsub subtree is mapped to a big xml structure, collections make 
problems.

> Would this help?
>
> http://www.jabber.org/jeps/jep-0163.html
>   

Don't think so - but I don't really understand it.

publish node='http://jabber.org/protocol/tune'

looks like mixing namespaces and node name.



Bernhard



More information about the Standards mailing list