[Council] meeting agenda, 2006-03-23

[Peter Millard]

Matthew A. Miller wrote:
> Now back to JEP-0124:
> I'm generally opposed to such large changes to Draft JEPs (which this 
> appears to me to be the case), unless the changes fix deficienies (e.g. 
> JEP-0045 and JEP-0060).  The biggest reason is that it makes it hard for 
> multiple implementors to properly interoperate, because they need to 
> somehow determine what version of a JEP the other parties have 
> implemented (unless the changes can be 100% ignored).
> 
> I'm wondering if the multi-stream stuff could not have been done as an 
> add-on (and therefore a different JEP), but I'll read through it and 
> weigh in later.

Hope you guys don't mind the intrusion :) I'm wondering if Ian could 
send out a link to standards-jig list thread which provides the 
background on this multi-stream stuff.

FWIW, Jabber, Inc. is building a -124 implementation, so I'm very 
concerned about these changes (it takes us a while to roll out changes 
after we've implemented). Upon first glance, it seems these changes are 
fine with previous versions of the JEP and should not pose any 
compatibility issues. What I don't understand is the problem that this 
new stuff is trying to solve. With -124, the client can maintain 
multiple connections to the CM, so why do we need multiple streams? It 
also seems that multiple streams just complicate matters with the 
in-order delivery requirements of RFC 3920 as well. IE, one stream can 
receive stanzas before the other stream and breaking in-order delivery.

pgm.