[ejabberd] About current pubsub implementation

Jehan Pagès jehan.marmottard at gmail.com
Mon Jun 16 17:08:07 MSD 2008


I have some questions about choices taken for the current pubsub
implementation. It ( http://www.ejabberd.im/mod_pubsub-usage ) is said:

Pubsub nodes are organized in a hierarchical tree, similar to a file system.
mod_pubsub imposes a special structure on this tree. There is a special
node, pubsub/nodes, that contains all other created nodes. Nodes of users
have names like home/*server*/*username*/*whatever*. If you try to create a
node whose name doesn't fit that pattern, you will get an error. If you
don't specify a name (i.e. create an *instant node*), the server generates a
name that fits this pattern.

So I have 2 questions:

1/ Why force people to make nodes of the form home/server/username/... This
would make sense for PEP, as it is "Personal". But for the generic pubsub
implementation, it does not make sense in my own opinion. I don't see why a
publication node MUST be associated to a specific login. First you may want
that people do not know your personal jid and still provide them with a
publication node. Or simply a node can be used for publication by
multi-publishers. A common example would be a general website with many

Anyway I don't think this is good. You may provide the Jabber server
administrators with a way to restrict the node names if they want. But I
really don't understand why it is done here at the software level (so no
other configuration).

2/ What is this special node pubsub/node? It says it containes "all other
created nodes". But if I create a node in my /home, I don't see it in
pubsub/nodes. And I cannot create anything in pubsub node. So what is it

For info, I use the last released version of ejabberd: 2.0.1.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jabber.ru/pipermail/ejabberd/attachments/20080616/b4c9540d/attachment.htm 

More information about the ejabberd mailing list