[jdev] Re: bytestreams and Psi

[Justin Karneges]

On Thursday 10 November 2005 02:07, Mats Bengtsson wrote:
> Sorry, I missed the last part here:
> "To avoid a race condition, the stream to use is to be decided by the
> original initiator" ------------------

Yes, that's an important bit. :)

> I think I will give it a try anyway.
> Does anyone the objections against this method to resolve the situation
> when the initiator in nonreachable (behind NAT)?
> Has this method been rejected formally by the Jabber guys?

The Fast Mode method has not been formally rejected, but the Role Reversal 
method has been.

Role Reversal worked by having the entities switch roles (Target would become 
the Initiator) after a failed S5B negotiation to try again in the other 
direction.  This feature was actually present in JEP-65 when it was advanced 
to Draft, but sometime later the council determined that this capability was 
too complex and it was removed from JEP-65.  I believe this was based on the 
results of a coding session involving Exodus and Proxy65.

Given that Role Reversal is much simpler than Fast Mode, yet was rejected, I 
have no plans to push Fast Mode in the JSF.  I did bring it up on the mailing 
list at one point as a suggestion, but that is all.

-Justin