[JDEV] Namespaces vs. Attributes

[Kriggs]

David Waite writes:
> In short, I don't think it is really anything that can be cleverly
> hacked in. Either a lot of things will break, or a lot of logic will
> need to be added for supporting legacy clients/servers. It has been off
> and on the roadmap for the open-source server, but I don't believe it
> will ever be changed/fixed until someone who actually cares decides to
> change/fix it.

The only trouble I could see is with the server keeping track of yet another 
variable. Still, I think that it would be worth it to work on the servers to 
incorporate it. The sooner, the less reprecautions. If there are side-effects 
now, imagine 5 years down the line. Jabber has been increasing steadily in 
popularity, and could very well be the primary IM of the future. The 
important thing is to get the servers compatible with this, and keep the 
clients doing what they've been doing so far. As servers upgrade, clients can 
slowly start taking advantage of the new feature.

Implementing this feature will actually be pretty safe. After implementation, 
the server will act the same way sending wise, so it won't confuse anyone, 
but will allow for the use of xmlns within the inner tags, which is not a 
useless idea, especially if Jabber really wants to conform to the XML 
standard.

If I were keeping track of the way the servers work, I'd probably put it on 
my "To do eventually" list. Some people could probably use some of my code 
when I'm done with my client, and mold it for the purposes (and language) of 
the server software: it treats the whole of the communication with the server 
as one document instead of many, meaning 'xmlns's can be declared anywhere, 
and they will be treated appropriately. (It'll be in Python).

-Kriggs
RBJab