[standards-jig] S5B Extension

Peter Saint-Andre stpeter at jabber.org
Thu Dec 11 21:19:22 UTC 2003


You have raised a policy issue, which needs to be investigated. This
email does not contain the results of such an investigation, since doing
so will involve reviewing list archives and chatroom logs in order to 
fully reconstruct the chronology of events, which will take some time. 

The issue of role-switching was heavily discussed on the lists, and IIRC
discussed on the Council. The revision you refer to was made during
OSCON while several people (among them a number of Council members) were
present and working to implement the JEP as defined. A number of changes
and improvements were made to the protocol definition at that time based
on real-time development experience, including changes to the zeroconf
handling. The removal of the role-switching feature may or may not have 
been a result of that experience -- I can't quite recall at this point,
since the changes were made at late-night hacking sessions for the most
part. The modified JEP (I believe Version 1.1) was published in July and
there has been ample opportunity for review and feedback since that time.

As to your statement that removing role-switching was "unfair", that is
your assertion. We need to investigate whether your statement is true.

As to your statement that these changes did not have to be seconded by
the members of the JSF, that statement is false, since JEP changes never
have to be seconded by the JSF membership. The process is clearly
defined in JEP-0001. However, as noted, the changed document was 
published on the jabber.org website and announced on this list.

As to your statement that the changes did not have to be reviewed by the
Council, I note that a good chunk of the Council was at OSCON actively
implementing the JEP and suggesting changes based on real-world
experience, and other members of the Council were being consulted on the 
issues in groupchat rooms or private chats. Before this protocol is
advanced to Final, the Jabber Council will of course review it again in
the light of development experience. That too is part of the process as
defined in JEP-0001.

You assume that the authors seem reluctant to add role-switching back.
That may or may not be the case. You sent a message about this on
2003-11-30. The fact that you have not yet received a reply does not
necessarily indicate that the authors will not make this change.

Therefore your decision to add a custom extension seems premature, and
in addition seems to violate the spirit of the JEP process. What if
every developer who based on experience did not like some aspect of a 
JEP decided to publish a personal extension rather than work with the
community to modify the JEP itself? That doesn't seem helpful to me.

However, as mentioned, I will perform a more complete investigation of
this matter when time allows. I will try to do that next week but I have
a lot of other commitments at this time related to our work in the IETF.


Peter Saint-Andre
Jabber Software Foundation

On Thu, Dec 11, 2003 at 01:37:44AM -0800, Justin Karneges wrote:
> As I reported earlier, role-switching was removed from JEP-65 sometime after 
> it advanced to draft.  This is unfair, I think, as these changes didn't have 
> to be seconded by the members or reviewed by the council.  Certainly a JSF 
> policy change is in order.
> Because the authors seem reluctant to put the role-switching feature back, I 
> have decided to create a custom extension to support it.
> Also, after doing some real-world tests with JEP-65, I realized it would be 
> useful to allow simultaneous connections from both peers, to avoid having to 
> wait for role-switches to happen.  I have called this 'fast-mode.'
> I have documented both extensions here:
>   http://affinix.com/~justin/stream.html
> The extensions were designed with backwards compatibility in mind.  Any client 
> supporting them should have no trouble communicating with those that don't.  
> I will not be pursuing a JEP.  I only mention them here because I would like 
> comments for improvement.
> -Justin
> _______________________________________________
> Standards-JIG mailing list
> Standards-JIG at jabber.org
> http://mailman.jabber.org/listinfo/standards-jig

More information about the Standards mailing list