Sorry, I gave some more thought to this after I wrote that email and I
think your proposal is better. But I haven't had time to post about it
in more detail yet...

On 2/13/10 7:40 AM, Waqas Hussain wrote:
> On Sat, Feb 13, 2010 at 10:43 AM, Peter Saint-Andre <[email protected]
> <mailto:[email protected]>> wrote:
> 
>     On 2/5/10 10:39 AM, Waqas Hussain wrote:
> 
>     > What exactly is the advantage of keeping the old <status/> element?
> 
>     Backward compatibility.
> 
> 
> What exactly does that mean? Does it lead to reduced code complexity on
> the server or on the client?
>  
> 
>     Is your proposal that services would send both
>     the old <status/> element(s) and your proposed <mucstatus/> element
>     (with potentially multiple children)? I don't see many benefits to that.
> 
> 
> The main reason I'm opposed to the old <status/> element is that the
> <status/> element can occur multiple times. Iterating over the children
> of all <status/> elements leads to awkward edge cases. This is increased
> code complexity on the client.
> 
> Consider:
> 
> <status xmlns='jabber:client'><ns1:child1/><ns2:child2/></status>
> <status xmlns='jabber:client'><ns3:child3/></status>
> 
> How should a client interpret the above? Does it take one child from
> each <status/> element? Which one? Does it merge them? And note that
> these children are in a different (possibly unknown) namespace;
> forbidding multiple such children in the schema would be even more
> awkward (not to mention against the spirit of namespaces).
> 
> And then there's the aesthetic reason. The following
> 
> <status>
>   <child1/>
>   <child2/>
> </status>
> 
> seems more readable to me than
> 
> <status>
>   <child1/>
> </status>
> <status>
>   <child2/>
> </status>
> 
> In the end, despite writing all of the above, I'm not actually too
> strongly opposed to doing it either way.
> 
> --
> Waqas Hussain
> 


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



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

Reply via email to