[standards-jig] JEP-0079 - Some Thoughts

Matthew A. Miller linuxwolf at outer-planes.no-ip.com
Mon May 5 17:19:54 UTC 2003

Just so y'all don't assume JEP-0079 (Message Delivery Semantics) is
dead, here's some thoughts on the matter from the author (-:

"RETURN RECEIPT" (aka "Delivery Confirmation")
I still feel this is not something for this JEP to tackle.  Delivery
means different things to different applications, and (to me) seems
better handled by another JEP that could expand and accommodate
some/most of those views.  For instance, answering the question "what
does it mean if a message 'times out'?" for a client can be answered in
many different ways, depending on what "client" means, and what that
client's users' perceptions are.  It could mean that the window never
got displayed, or that the user didn't give said window focus, among
others.  On the other hand, the answer to the same question for a server
(to me) is very clear: it means the message did not leave this server
for the next point in time.

As currently defined in the "published" revision, "expire-in" is
ambiguous.  I've changed this to be more absolute, in that each
processor decrements this value by the amount of time it takes them to
process, and meets the condition if the value is less than 0.  The term
"process" needs to be defined, but otherwise this should suffice.

Due to (un) popular demand, all "match-resource" condition values have
been remove, with the exception of "exact".  Once y'all can agree on
what other possible condition values can exist (and what they mean),
they can be included.

This is an issue that Marshall Rose pointed out.  As currently defined,
JEP-0079 does not take intermediate servers into account.  Some
suggestions that have been made for this include:

1)  Add a "must-process" attribute to each <rule/>, specifying whether
the outer-most servers must process this rule ("firstlast") or every
server in the route must process this rule ("each").  Additionally, each
rule definition specifies whether this flag applies (e.g. condition
"match-resource" does not apply, as only the last server can determine
2)  Include a <stream:features/> for announcing JEP-0079.  This feature
is to ensure that each hop in the route used to deliver the message can
handle the processing rules.  The specific details of this added feature
are TBD.

Comments, questions, thoughts?


Matt "linuxwolf" Miller
JID:	linuxwolf at outer-planes.net
E-MAIL:	linuxwolf at outer-planes.net

- Got "JABBER"? (http://www.jabber.org/)

More information about the Standards mailing list