On 12/3/09 12:54 PM, Robin Collier wrote:
> 
> 
>> Date: Thu, 3 Dec 2009 11:56:30 -0700
>> From: [email protected]
>> To: [email protected]
>> Subject: [PubSub] finishing up 1.13
>>
>> OK folks, it's time to push version 1.13 of XEP-0060 out the door. My
>> main question is: do we have consensus on the versioning stuff? If not
>> (and if we can't gain consensus soon), then I would vote to remove it so
>> that we can publish 1.13.
>>
> 
> I am sure it comes as no surprise, but I believe 'ver' values need to be
> ordered 

In roster versioning (XEP-0237), we said only that the 'ver' needs to be
unique, leaving it up to the implementation whether it will make the
'ver' a strictly-increasing sequence number (and I think that's the
consensus for pubsub, as well). It's not clear to me why pubsub 'ver'
needs to be ordered. Could you please recap your argument for why *all*
pubsub implementations (not just your implementation) MUST enforce ordering?

> and present in the request for items on a per item basis 

A request for items is (typically) a request for 1+ items, and that's
covered by a new section in the spec. A client is free to include 'ver'
on the request, but is not required to do so. Are you suggesting that
the spec makes this a MUST? Do you also think that a request for a
particular item MUST include the 'ver'?

> to resolve
> the issue with synch and async retrieval of the same item at the application
> level.  This would then resolve more use cases than the opaque id.

So implement it that way in your system. I still don't see why this
needs to be a MUST for all implementations.

Peter

-- 
Peter Saint-Andre
https://stpeter.im/


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to