[Council] proto-JEP: Stream Acks

Robert Norris rob at cataclysm.cx
Wed Aug 11 18:36:29 CDT 2004

On Tue, Aug 10, 2004 at 12:32:58PM -0500, Peter Saint-Andre wrote:
> Justin Karneges has submitted a proposal for acknowledgement of stanzas:

I don't object quite as vociferously as the rest of you for the simple
reason that this JEP looks more aimed at doing acknowledgement between
two stream endpoints rather than two entities. This distinction is
important - a stream endpoint is rarely a fully addressable entity (eg a
client and a c2s component - both entities involved (the user and the
"server") are actually held at the session manager). In this case,
normal acknowledgements (via some sort of tagged reply packet) aren't
necessarily the right thing to do.

Adding "toplevel" elements is not an issue IMO, since they're done in a
different namespace, and triggered via a feature - exactly the same as
how SASL, STARTTLS, etc work. Since its a stream level thing, there's no
reason that it couldn't be doing acknowledgements for other namespaces
(ie not jabber:client) - not using iq/presence/message in this case is
probably a good thing.

As it states, the ping is nicer than simply sending whitespace (though
it is heavier) - the ping allows the client to determine if the server
is still there, as opposed to simply forcing the connection to stay open
(connection being open doesn't always mean that the server is still

I think there needs to be more stuff to tie an acknowledgement to the
packet that triggered it, as well as text about what to do if the ack
isn't received - handling timeouts, resends, etc. This is particularly
important if the acks are supposed to be used to determine if throttling
is in place. For checking for throttling, I think I'd prefer a seperate
protocol where the server informs a client when its being throttled, and
when the throttling is over.

I think the most useful thing this JEP could gain before being published
as Experimental is a clear reason for why its needed. Perhaps with an
example or two. Apart from that, I see no reason to hold it back.


Robert Norris                                       GPG: 1024D/FC18E6C2
Email+Jabber: rob at cataclysm.cx                Web: http://cataclysm.cx/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://jabber.org/pipermail/council/attachments/20040812/888e9a10/attachment.pgp

More information about the Council mailing list