[Standards-JIG] Stanza Acking and Connection Pings

Pavel Šimerda pavel.simerda.lists at centrum.cz
Fri May 19 12:21:05 UTC 2006


Hi people

There was a long debate about stanza acking and connection (per-hop) 
reliability.

1) There was a fine proposal from Justin Karneges that solved the problem very 
simply. It can be seen at http://www.jabber.org/jeps/inbox/ack.html .

2) Then there was a complex solution from Dave Cridland... which included new 
things like numbering the stanzas to handle reconnection directly... 
differenciating between unsolicited acks inside subsequent stanzas... and 
solicited acks, got using <iq> stanzas.

For both of the proposals, timeouts could be negotiated by the clients... 
because users usually know better if they're on a bad connection than 
servers. There are two types of timeouts, one for message to be acked and one 
for silent connection to be tested/pinged (a longer one).

We already know there would be a good number of users who would benefit from 
making xmpp communications more reliable. And making the use of jabber more 
comfortable, of course.

The two crucial points are... to know that the message got through the 
connection and the program is no longer responsible for it... and to know 
that the connection doesn't work any more and reconnect (if desired). This 
works with both proposals.

Alright, what's going on now? Could we re-join forces split between Justin's 
and Dave's proposals? And have something finished?

Pavel



More information about the Standards mailing list