[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [Sip] INFO considered harmful
I think people are suffering from the delusion that because SIP
has already established a channel of communication between the
transport layer of the SIP entities on the involved machines, that
it then becomes "easier" to get information to the application layer
by tunneling it through INFO on this established channel, and that
you gain something by the "handshakes" provided "free" by SIP.
In fact, nothing at all has been gained. All of your application
connection problems remain. Michael is right in saying it provides
no such benefit, and it makes NO difference from an app perspective,
if you tunnel junk through INFO or roll you own protocol over
another port.
Which is not to say INFO isn't useful.
Gordon
> -----Original Message-----
> From: Michael Thomas [mailto:mat@cisco.com]
> Sent: Thursday, January 02, 2003 12:53 PM
> To: Eric Burger
> Cc: Dean Willis; Jonathan Rosenberg; sip@ietf.org
> Subject: RE: [Sip] INFO considered harmful
>
>
>
> I don't get it. If you want "100% application
> specific", why not pick an unused port and roll
> your own protocol? Then all of the crufty SIPism
> don't get in the way, and the hue and cry from
> this working group ceases immediately. Do the bits
> on the wire keep warmer if they're dressed in SIP?
>
> Mike
>
> Eric Burger writes:
> > I fully agree with Dean on this one.
> >
> > One point to consider (I'm on the fence on this one) is
> whether we would say that INFO is 100% application-specific.
> That is, the WG will not publish standard body types.
> >
> > A reason for doing that is to let people know that INFO
> really is just, as Dean points out,
> application-to-application communication. If you want
> applications to interoperate, take it either to an
> Application Area WG or take it outside the IETF.
> >
> > One reason to not do this is SIP-T should use a method
> other than INFO. That said, I don't think there are that
> many SIP-T implementations extant.
> >
> >
> > > -----Original Message-----
> > > From: Dean Willis [mailto:dean.willis@softarmor.com]
> > > Sent: Tuesday, December 31, 2002 1:48 AM
> > > To: Jonathan Rosenberg; Orit Levin
> > > Cc: sip@ietf.org
> > > Subject: Re: [Sip] INFO considered harmful
> > >
> > >
> > > At 12:48 AM 12/31/2002 -0500, Jonathan Rosenberg wrote:
> > > >Orit Levin wrote:
> > > >>>The big huge difference I have been trying to point out is
> > > that EVENTS
> > > >>>DEFINES SEMANTICS, whereas INFO doesnt.
> > > >>>MIME provides semantics and syntax that are independent of
> > > the type.
> > > >>>SIP events provides semantics and syntax that are
> > > independent of the
> > > >>>package.
> > > >>>INFO provides nothing.
> > > >>
> > > >>INFO provides "minimal" but very important semantics:
> > > asynchronous data
> > > >>reliably follows the established SIP path.
> > > >
> > > >Huh?
> > > >
> > > >Every new SIP method is reliable, it inherits the SIP
> > > transaction state
> > > >machine for non-invite. Follows the established path? Every
> > > in-dialog
> > > >method would have this property, it is method independent.
> > > >Asynchronous? You can send any method at any time.
> > > >
> > > >Thus, nothing you have pointed out above is different
> > > between INFO and any
> > > >other new method we might introduce.
> > >
> > > I think the key point is that there is a difference
> between transport
> > > protocol semantic and application semantic. The transport
> > > protocol semantic
> > > expressed in INFO is "Here is some data that will be used by the
> > > application. It is not important to the transport protocol,
> > > except that the
> > > transport protocol is expected to deliver it reliably". INFO
> > > says NOTHING
> > > about the APPLICATION level semantic -- that's up to what
> > > goes IN the INFO
> > > payloads. The question is -- do we try to rigidly define
> > > application-level
> > > semantics here? Do we define that there ARE no such
> > > applications possible?
> > > Or do we design a framework by which application-level
> > > semantics can be
> > > expressed outside of the protocol definition?
> > >
> > > And as for "Just add another method to your SIP stack". Let's
> > > say I have a
> > > mobile phone and the SIP stack is burned into ROM and
> exposes only a
> > > simplistic transactional API. This is only likely to
> happen, oh, 100
> > > million times or so over the next two years . . . Now, just
> > > exactly how is
> > > my BREW or JTME application going to go about extending the
> > > SIP stack to
> > > support another method? Will the evil wireless operator
> > > networks even pass
> > > another method? Neither is happening anytime soon, I think . . .
> > >
> > > --
> > > Dean
> > >
> > >
> > > _______________________________________________
> > > Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
> > > This list is for NEW development of the core SIP Protocol
> > > Use sip-implementors@cs.columbia.edu for questions on current sip
> > > Use sipping@ietf.org for new developments on the
> application of sip
> > >
> > _______________________________________________
> > Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
> > This list is for NEW development of the core SIP Protocol
> > Use sip-implementors@cs.columbia.edu for questions on current sip
> > Use sipping@ietf.org for new developments on the application of sip
> _______________________________________________
> Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
> This list is for NEW development of the core SIP Protocol
> Use sip-implementors@cs.columbia.edu for questions on current sip
> Use sipping@ietf.org for new developments on the application of sip
>
_______________________________________________
Sip mailing list https://www1.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use sip-implementors@cs.columbia.edu for questions on current sip
Use sipping@ietf.org for new developments on the application of sip