[Standards] [Standards-JIG] XEP-0124 Binding->BOSH

Ian Paterson ian.paterson at clientside.co.uk
Thu Feb 15 05:45:03 UTC 2007


Hi,

In the interests of perhaps proposing XEP-0124 as an IETF standard, I've 
split it into two documents. XEP-0124 is now a generic XML streaming 
protocol (renamed to "BOSH"), while a proto-XEP (nominally called 
XEP-0206 - XMPP over BOSH) includes a couple of XMPP-specific extensions 
and a clarification of the best practices.

These documents are complete rewrites, but almost all the changes were 
for clarity (XEP-0124 wasn't a nice document). The few changes are 100% 
backward compatible, with the possible exception of the xmpp:restart 
attribute in XEP-0206 which enables SASL auth (SASL was not properly 
supported properly under the previous version of the protocol).

Two exciting new features are the "Script Syntax" which enables browsers 
to make cross-domain connections, and "session pauses" which enable 
portlets.

If you never quite understood how the protocol should be used then 
XEP-0124's new section 4: "The BOSH Technique" should make things clear.

You can look at the HTML versions here:

http://www.clientside.co.uk/xeps/xep-0124/xep-0124.html

"Multiple clarifications and restructuring without changes to protocol 
itself; changed title to BOSH; added section that fully explores the 
technique underlying the protocol; separated XMPP-specific features into 
new XEP-0206; added optional new Script Syntax and session pauses; added 
from and ver attributes; added hold attribute to session creation 
response; clarified polling too-frequently error; recommended clients 
use HTTP Pipelining."


http://www.clientside.co.uk/xeps/xep-0206/xep-0206.html

"Initial version (extracted from XEP-0124 version 0.15); depricated 
non-SASL authentication and authid attribute; multiple clarifications 
and restructuring without changes to protocol itself; added optional 
version and restart attributes."


I'm going to start work on BOAH now. It's similar to BOSH but based on 
the Comet technique.

- Ian




More information about the Standards mailing list