[Standards] BOSH (was HTTP Binding) namespaces

Ian Paterson ian.paterson at clientside.co.uk
Sun Mar 4 13:03:56 UTC 2007


Alex Gnauck and Mridul have independantly pointed out the fact that all 
XEP-0124 implementations (including my own) do not add "jabber:client" 
namespace declarations to the stanzas they pass through (despite the 
fact that all the examples in XEP-0124 include the namespace). This is 
for a very good reason, since the connection manager is then able to 
forward everything transparently between the client and server. However 
the result does not conform with the schemas (or examples).

To allow transparent forwarding while staying true to the schemas, I'm 
considering editing the schemas and the examples in XEP-0124 and XEP-0206:

1. The schema defined in XEP-0124 will be renamed from 
'http://jabber.org/protocol/httpbind' to 'urn:xmpp:bosh' and all the 
examples will be changed accordingly.

2.The schema defined in XEP-0206 will be renamed from 'urn:xmpp:bosh' to 
'http://jabber.org/protocol/httpbind' and I will *copy and paste* the 
content of the 'jabber:client' schema and the content of the schema 
defined in XEP-0124 into that schema!

Therefore all the stanzas in the examples in XEP-0124 will include 
"jabber:client" default namespace declarations, but these will be 
omitted from the examples in XEP-0206 since everything will be covered 
by the default namespace. In addition the "xmpp:" prefix will be omitted 
from the examples in XEP-0206.

These changes will ensure that all existing (and future) implementations 
magically become fully compliant with *XEP-0206* (even if not with the 
schema defined in XEP-0124).

In the future the schema in XEP-0206 will need to be updated to reflect 
any changes to the 'jabber:client' schema or to the schema defined in 
XEP-0124, but those are likely to be stable.

Finally, I could swap the names of the two documents, so any references 
to XEP-0124 or literature that claims an implementation is XEP-0124 
compliant do not have to be changed.

What do implementors think?

- Ian




More information about the Standards mailing list