On 07/08/2011 11:51 PM, Justin Karneges wrote:
> On Friday, July 08, 2011 04:44:20 AM Sergey Dobrov wrote:
>> On 07/08/2011 04:37 AM, Justin Karneges wrote:
>>>   b) if RSM is used by the server, then items are returned in a
>>>   node-specific
>>>
>>> order.
>>
>> What means "node-specific"?
> 
> I just mean, the ordering would be defined by the node type or node 
> implementation.  For example, XEP-0277 (microblogging) might want to say that 
> if RSM is used, then items are ordered by modified time ascending.  However, 
> other nodes defined in other XEPs might have different orderings based on 
> criteria other than dates.
Sounds good but there are no nodes that have different behavior based on
it's content. Suddenly, node is more low-level term.

> 
>>>   d) if neither RSM nor max_items are used, then items are returned in
>>>
>>> unspecified order.
>>
>> Not all payloads can be sorted on client side. I think that we need to
>> return some metadata too including creation and modification times as
>> well as item publisher.
> 
> Ah probably.  I think this is a separate issue from RSM though.
Right. But any issue that I trying to discuss about is silently ignored
here. I don't know why.

> 
>> Add a "Item ordering" paragraph to XEP-0060?
> 
> I think we just need to complete the description of RSM.  There is a strong 
> feeling, though, that we should not add to XEP-0060 since it is already too 
> long.  However, this dangling underspecified RSM in the spec is a bit odd 
> IMO.  
> Here are the options I suggest:
> 
> 1) Add remaining description of RSM into XEP-0060.
> 
> 2) Remove RSM from XEP-0060 (the XEP gets shorter!) and create a separate XEP.
Agreed.

> 
> Justin
> 


-- 
With best regards,
Sergey Dobrov,
XMPP Developer and JRuDevels.org founder.

Reply via email to