Hi, 

>>Is this really a problem for non-INFO message bodies? There is no CID
for SDP, but clients seem to be able to deal with it anyway.
>
>There is no CID, but there is a C-D: an implicit C-D of "session" for
app/sdp, and "render" for everything else. 

Sure. But, assume we would also be using SDPng. It would also have a C-D
value of "session", but I assume the C-T value would be different than
"app/sdp". You need to know what is in the body, so you can process and
parse it correctly.

>And I am not at all advocating that we use CID for the body-parts that
actually belong to 
>the message context.  Quite the contrary, I think it would be silly to
do that.
>
>And yes it is a potential problem for non-INFO message bodies, in
exactly the same way it's a potential problem for INFO.  There is
nothing unique about INFO which makes the problem only applicable to 
>it.
>
>To reiterate the "problem": the issue is not that messages can have
multiple body-parts.  The issue is that one or more body-parts may not
actually be intended for the message's context.  In other words, 
>they're not intended for the application-layer processing logic
specific to that method (and package if there is one).  For example,
when I send a SUBSCRIBE request, the method name "SUBSCRIBE" and the 
>Event header package name identify a specific context -  a specific
use-case of the message with its own set of rules and for a specific
application-layer above, which actually understands that package's 
>specific semantics and how to process the higher-level "subscription".
>
>So if the SIP message has body-parts, and one of them is actually not
intended for that package but is instead just a generic add-on, like
geoloc for example, we need to dis-ambiguate that add-on body-
>part(s) from the others.

It should be defined in what SIP messages geoloc information may be
transported, and what the receiver supporting it is supposed to do with
it. As we do for SDP. SDP in CANCEL has no meaning, for example, so it
is seen as a protocol error to send SDP in CANCEL.

I see no reason why we should solve things so that one can insert
whatever, wherever...

And, as far as INFO is concerned, I guess we all agree that info
packages are within the context of INFOs. BUT, you still need to know
body part(s) contain info package bodies, and what info package is
containded in a specific body part. C-T can do that, as it does for XML.

Regards,

Christer
_______________________________________________
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

Reply via email to