On Fri, Dec 16, 2011 at 12:22 PM, Kevin Smith <[email protected]> wrote:
> On Thu, Dec 15, 2011 at 5:22 PM, Matthew Wild <[email protected]> wrote:
>> On 15 December 2011 15:27, Matthew A. Miller <[email protected]> 
>> wrote:
>>> On Nov 1, 2011, at 15:36, Kim Alvefur wrote:
>>>
>>>> Hi!
>>>>
>>>> I've noted that Message Forwarding [XEP-0297] doesn't give any
>>>> recommendations for how (or if) you should indicate to which extension
>>>> it belongs.  Is this something that would be desirable?
>>>>
>>>> Currently Message Carbons [XEP-0297] has what can be seen as an
>>>> indicator inside the <forwarded/> element[1].  Would it be better or
>>>> worse to, let's say, have this as a child of the parent <message/>?
>>>>
>>>> 1: http://xmpp.org/extensions/xep-0280.html#example-13
>>>
>>> To be honest, I don't remember why I chose to put the indicators as 
>>> children of the <forwarded/> instead of as siblings.  I can see either 
>>> pattern making sense, and am happy to change XEP-0280 if the current 
>>> pattern is too strange.
>>>
>>> Maybe one of XEP-0297's authors can comment on this?
>>>
>>
>> I'm inclined to say it should be a sibling. When you consider how an
>> application is going to sensibly handle this, it's going to first want
>> to categorize it as a carbons message before extracting the message
>> contents from <forwarded/>. Partly the principle is that you can judge
>> a message type by the namespaces of its immediate children.
>
> I agree.

I have changed my mind, as indicated in the other thread. I think the
use of a forward should be a child of the using protocol. So for
carbons, you'd have:

<message><carbon><forward><message/></forward></carbon></message>

The forward is a property of the carbon, not a property of the message
(Using the 'can use direct children to determine type of parent' idea
- Message "is a" Carbon, Carbon "is a" forward)

/K

Reply via email to