Re: Proposal for multiple INFO bodies in an INFO

Hi Erick,
 
I am not your suggestion of using CID for Info packages is what has been discussed (I may be wrong, of course).
 
I think what Paul was proposing was to have a Content-Disposition value indicating that a particular body shall be treated as in info package.
 
INFO .....
To: ....
From: ....
Info-Package: foo
Content-Type: application/foo
Content-Dispotistion: info
...

<foo body>


Multipart mixed example with indirect content:

INFO ....
To: ....
From: ....
Info-Package: foo
Content-Type: multipart/mixed;boundary="theboundary"
...

--theboundary
Content-Type: application/mumble
Content-Disposition: whatever

<mumble body>

--theboundary
Content-Type: application/foo
Content-Disposition: info

<foo body>

--theboundary--

 
 
Regards,
 
Christer

________________________________

From: [email protected] on behalf of Eric Burger
Sent: Thu 30/10/2008 19:29
To: IETF SIP List
Subject: [Sip] Proposal for multiple INFO bodies in an INFO



Good thing I waited 10 days...

Initially, I was very swayed by the argument that things would be 
much, much easier if we restricted an INFO request to a single Info-
Package.

However, and especially as the last person to touch RFC 4483, I can 
see the argument that *any* SIP message can end up with a multipart/
mime body, even if you think you can limit it to just your stuff, 
because the UA can always have a content indirection pointer in the 
header pointing to a mime body part.

Therefore, even if we restrict an INFO request to carry a single Info-
Package, we need a mechanism to uniquely identify and, at the SIP 
header level, indicate which body part belongs to the package.

I would offer once we solve that, we also get multiple INFO request 
bodies for free.

I would also offer that since INFO does not change SIP state, which 
means there is not even a concept of a message succeeding but the INFO 
body doing something that "fails", the objection that it would be hard 
for a UA to report on one body "failing" with another body 
"succeeding" is a non-issue.  Yes, this does mean that you cannot use 
INFO to tunnel IP.  See RFC 3252 for more on this.  I do not see this 
as a problem.

Therefore, on to the solution.  We use the MIME approach.  The body 
parts get tagged with Content-IDs. The Info-Package has a parameter, 
CID, which points to the body part.  The ONLY question I have is 
whether we should simply mandate always having this parameter to make 
the protocol and parser easier, but a small burden on UAC's that 
really are only sending a single body.  Other than that, the mechanism 
looks like:

INFO .....
To: ....
From: ....
Info-Package: foo
Content-Type: application/foo;cid=this
Content-Id: this
...

<foo body>



Multipart mixed example with indirect content:

INFO ....
To: ....
From: ....
Info-Package: foo;cid=abcd1234zz
Mumble: <cid:abcd9999qq>
Content-Type: multipart/mixed;boundary="theboundary"
...

--theboundary
Content-Type: application/mumble
Content-Id: abcd9999qq
...

<mumble stuff>

--theboundary
Content-Type: application/foo
Content-Id: abcd1234zz

<foo body>

--theboundary--




Multiple Info Packages follow. Note that by using Content-Id, order is 
irrelevant.

INFO ....
To: ....
From: ....
Info-Package: foo;cid=abcd1234zz
Info-Package: bar;cid=wxyz9876
Content-Type: multipart/mixed;boundary="theboundary"
...

--theboundary
Content-Type: application/bar
Content-Id: wxyz9876
...

<bar stuff>

--theboundary
Content-Type: application/foo
Content-Id: abcd1234zz

<foo body>

--theboundary-- 

_______________________________________________
Sip mailing list  https://www.ietf.org/mailman/listinfo/sip
This list is for NEW development of the core SIP Protocol
Use [email protected] for questions on current sip
Use [email protected] for new developments on the application of sip