[Standards-JIG] JEP-60 Item id for node configuration change notifications

Ralph Meijer jabber.org at ralphm.ik.nu
Wed May 31 19:56:29 UTC 2006


On Tue, May 30, 2006 at 01:54:43PM -0600, Peter Saint-Andre wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Ralph Meijer wrote:
> > On Wed, May 10, 2006 at 01:14:34PM -0600, Peter Saint-Andre wrote:
> >> [..]
> >>
> >> I'm not sure why Peter did it that way, since that's the only reserved
> >> ItemID as far as I can see. However, I don't immediately see a better
> >> approach, so I think a service would in fact need to reject publication
> >> requests that specify an ItemID of "configuration" (i.e., reserve that
> >> ItemID for generation only by the service). I'll clarify that in the text.
> > 
> > I dislike reserved ItemIDs. 
> 
> I don't like them, either.
> 
> > Can't we add a <configure/> to <event/> in
> > the #event namespace?
> 
> Like this?
> 
> <message from='pubsub.shakespeare.lit' to='francisco at denmark.lit' id='foo'>
>   <event xmlns='http://jabber.org/protocol/pubsub#event'>
>     <items node='blogs/princely_musings'>
>       <configure/>
>     </items>
>   </event>
> </message>

No, the idea is to include the configuration in the event, because that
is what the reserved item id was intended for. So more like this:

  <message>
    <event>
      <configure node='blogs/princely_musings'>
        ... JEP-0004 form in here ...
      </configure>
    </event>
  </message>

> > Also, I'm still not sure about the access attribute to the <configure/>
> > element in the other namespaces. I know it would be easier for PEP, but
> > I don't know if avoiding forms there is needed/a good idea.
> 
> I don't see any great harm in the 'access' attribute, just as I don't
> see any great harm in the 'type' attribute in an example like this:
> 
> <iq type='set'
>     from='bard at shakespeare.lit/globe'
>     to='pubsub.shakespeare.lit'
>     id='create3'>
>   <pubsub xmlns='http://jabber.org/protocol/pubsub'>
>     <create node='announcements' type='collection'/>
>   </pubsub>
> </iq>
> 
> That is, we could do that with a data form that includes the correct
> value of the "pubsub#type" field. But do we gain anything by removing
> the 'access' and 'type' attributes here and using data forms instead?

Heh, yeah. I think we added those attributes to simplify the protocol
for PEP. For now I think we should only use forms, really. The 'problem'
is that we now have configuration data spread over attributes of
<create/> and the contents of <configure/> (whether as JEP-0004 form or
some other namespaced XML for custom implementations).

-- 
Groetjes,

ralphm



More information about the Standards mailing list