[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Simple] RE: Advanced IM requirements



Hi,

See inline:

> -----Original Message-----
> From: simple-admin@ietf.org [mailto:simple-admin@ietf.org]On Behalf Of
> ext Eric Burger
> Sent: 29 February, 2004 15:31
> To: Chris Boulton; Niemi Aki (Nokia-M/Espoo)
> Cc: simple@ietf.org
> Subject: RE: [Simple] RE: Advanced IM requirements
> 
> 
> The differences between the drafts goes much farther than 
> e-mail format versus XML format.  That said, other than "XML 
> is cool", is there a reason for using XML instead of reusing 
> the Message Disposition Notification's e-mail-oriented syntax?
> 
> First and foremost, the sipping-message-receipt draft does 
> not specify the reporting period.  simple-imdn explicitly 
> states that the recipient will send one and only one 
> notification.  This simplifies state management at the 
> sender.  It also eliminates an avenue for a denial-of-service 
> attack, whereby the sender gets flooded with notifications as 
> the message is received, processed, read, and deleted.
> 
> If there is a requirement for following the entire lifetime 
> of a message, SUBSCRIBE/NOTIFY is a much better technology 
> than either sipping-message-receipt or simple-imdn.  That 
> said, I cannot imagine a use case that requires such 
> notification.  Am I missing something?
> 

I guess SUBSCRIBE/NOTIFY still has problems over a long time period, as it requires refreshing, and the recovery of subscription state may be hard after e.g. powerdown, which is a quite normal thing with wireless endpoints.

Also, the notifications themselves, if they can be sent over a long time period, should be possible to store, if their recipient is not on-line. This seems to make MESSAGE a more suitable carrier.

Markus

> Paul's comments hold: GRUU is not likely to be a good idea 
> for notifications.  In fact, the reason that e-mail 
> notifications and simple-imdn notifications can go to a 
> different URI than the message sender is that the 
> notification processor is often an automaton.  The user does 
> not want to see the notifications -- they just want to see 
> the blinky light turn off or the grey icon turn solid.
> 
> Moreover, the UAC instance could be long gone by the time the 
> message has a disposition.  If you were in an interactive 
> conversation, you wouldn't need notification in the first place...
> 
> Is there a reason for the notification request and protocol 
> machinery to be a SIP, and not CPIM, construction?  Is this a 
> generic notification protocol for SIP?  If so, are there use 
> cases outside of CPIM?  This also is an issue for the report 
> MIME type.  It is OK if it is a generic SIP tool.  However, I 
> would call it something other than 
> application/receipt-info+xml if it is only for IM.  I would 
> call it what it is, something like 
> application/simple-notification+xml.
> 
> One place where the issue of notification being a SIP vs. 
> CPIM construct is message correlation.  The 
> sipping-message-receipt draft goes to great lengths (e.g., 
> invoking GRUU and GRID) to identify a message.  However, 
> there already is a MsgID parameter in CPIM.  Users have no 
> concept of GRUU or GRID, but they do have concepts (or at 
> least the user interface can make sensible indicators) of 
> message ID.  Likewise, the use of Call-ID for <message-id> 
> doesn't make sense.
> 
> The normative UAS behavior is, IMHO, not correct for a 
> message that requires disposition notification.  If the UAC 
> requires a message disposition report, and the UAS refuses to 
> give one, the message should be rejected.  In the e-mail 
> world, this is how we train UAC's to not ask.
> 
> On the processing of the Receipt-To header, the 
> sipping-message-receipt draft suggests that "Receipt-To" 
> SHOULD NOT be in a notification.  This really, really, really 
> has to be MUST NOT.  Under what circumstance is it OK for a 
> notification to ask for a notification?  While we're at it, 
> the UAC needs a "MUST disregard Receipt-To in a notification".
> 
> Concerning the <final-destination> element: It is useful to 
> expose the route choices for error reports.  However, for 
> notification reports is it a good idea to expose the UAS' 
> actual Request-URI?
> 
> simple-imdn explicitly did not include the <deleted> state.  
> What does that mean for an Instant Message?
> 
> > -----Original Message-----
> > From: Eric Burger 
> > Sent: Friday, February 27, 2004 4:24 PM
> > To: Jonathan Rosenberg; Chris Boulton
> > Cc: simple@ietf.org
> > Subject: RE: [Simple] RE: Advanced IM requirements
> > 
> > 
> > The IMDN proposal (draft-burger-simple-imdn-00.txt) reuses 
> > all of the protocol machinery, parsers, and deployment 
> > experience of the MDN machinery from the messaging world.
> > 
> > Do note that almost half of the IMDN draft deals directly or 
> > indirectly with security issues.  The message-receipt draft 
> > will need to be expanded greatly to address the IESG issues 
> > that are sure to arise that the messaging folks have hashed 
> > through in the past.
> > 
> > I'll take a closer look at the message-receipt draft over the 
> > weekend (isn't that what long plane flights are for?).  
> > Expect more on Monday, Korea Time :-)
> > 
> > > -----Original Message-----
> > > From: Jonathan Rosenberg [mailto:jdrosen@dynamicsoft.com]
> > > Sent: Thursday, February 26, 2004 1:51 AM
> > > To: Chris Boulton
> > > Cc: Miguel.An.Garcia@nokia.com; simple@ietf.org; Eric Burger
> > > Subject: Re: [Simple] RE: Advanced IM requirements
> > > 
> > > 
> > > 
> > > 
> > > Chris Boulton wrote:
> > > 
> > > > 1. it placed the URI for sending the delivery 
> > notification as a CPIM
> > > > header, not a SIP header
> > > > 
> > > > <<CJB>> Out of interest, what was the driving force for 
> > > this design choice?
> > > 
> > > I believe the document talks a bit about this. It would make the 
> > > mechanism applicable to any system that uses cpim, which 
> > > would include 
> > > IM sessions and in theory any other IMPP compliant IM 
> > > protocol, though 
> > > since XMPP is not using it there are practically none others.
> > > 
> > > Of course, Eric is the author of this and is better suited to 
> > > answer the 
> > > question.
> > > 
> > > -Jonathan R.
> > > -- 
> > > Jonathan D. Rosenberg, Ph.D.                600 Lanidex Plaza
> > > Chief Technology Officer                    Parsippany, NJ 
> > 07054-2711
> > > dynamicsoft
> > > jdrosen@dynamicsoft.com                     FAX:   (973) 952-5050
> > > http://www.jdrosen.net                      PHONE: (973) 952-5000
> > > http://www.dynamicsoft.com
> > > 
> > > 
> > 
> > 
> > _______________________________________________
> > Simple mailing list
> > Simple@ietf.org
> > https://www1.ietf.org/mailman/listinfo/simple
> > 
> > 
> 
> 
> _______________________________________________
> Simple mailing list
> Simple@ietf.org
> https://www1.ietf.org/mailman/listinfo/simple
> 

_______________________________________________
Simple mailing list
Simple@ietf.org
https://www1.ietf.org/mailman/listinfo/simple