[Standards] (ab)uses of 0059 in 0060

[Bruce Campbell]

Result Set Management (0059) defines a (draft) standard for limiting what 
is returned as a result of a query, and how many possible results there 
are.  In reading it, there are no suggestions as to how 'deep' within a 
protocol it can be used.

For Yet-Another-Creative-Overlay-To-PubSub XEP that I'm writing[1], I'm 
wanting to use RSM to partially retrieve a given item, and by extension, 
to use RSM to allow <publish/>ing a given item in piecemeal fashion.

For example, instead of publishing an item as:

  <iq type="set" to="pubsub.example.com" from="someone at example.com" id="create0">
   <pubsub xmlns="http://jabber.org/protocol/pubsub">
    <publish node="foo">
     <item id="bar">
      <data>a big chunk of data (1)</data>
      <data>another big chunk of data (2)</data>
      <data>really big chunk of data (3)</data>
     </item>
    </publish>
   </pubsub>
  </iq>

I'm wanting to use a piecemeal and possibly out-of-order upload of:

  <iq type="set" to="pubsub.example.com" from="someone at example.com" id="create1">
   <pubsub xmlns="http://jabber.org/protocol/pubsub">
    <publish node="foo">
     <item id="bar">
      <data>really big chunk of data (3)</data>
      <set xmlns="http://jabber.org/protocol/rsm">
       <index>3</index>
      </set>
     </item>
    </publish>
   </pubsub>
  </iq>

  <iq type="set" to="pubsub.example.com" from="someone at example.com" id="create2">
   <pubsub xmlns="http://jabber.org/protocol/pubsub">
    <publish node="foo">
     <item id="bar">
      <data>a big chunk of data (1)</data>
      <set xmlns="http://jabber.org/protocol/rsm">
       <index>1</index>
      </set>
     </item>
    </publish>
   </pubsub>
  </iq>

  <iq type="set" to="pubsub.example.com" from="someone at example.com" id="create3">
   <pubsub xmlns="http://jabber.org/protocol/pubsub">
    <publish node="foo">
     <item id="bar">
      <data>another big chunk of data (2)</data>
      <set xmlns="http://jabber.org/protocol/rsm">
       <index>2</index>
      </set>
     </item>
    </publish>
   </pubsub>
  </iq>

And likewise a possible partial retrieval result:

  <iq type='result' from='pubsub.example.com' id='this_might_kill_my_karma'>
   <pubsub xmlns='http://jabber.org/protocol/pubsub'>
    <items node='foo'>
     <item id='bar'>
      <set xmlns='http://jabber.org/protocol/rsm'>
       <index>2</index>
       <count>3</count>
      </set>
      <data>another big chunk of data (2)</data>
     </item>
    </items>
   </pubsub>
  </iq>

Anyone with strong feelings one way or the other on this behaviour?

-- 
   Bruce Campbell

   [1] IMAP over XMPP, if you're curious.