[rohc] RE: [Sipping] P-Headers in the static SIP/SDP compression diction ary

If the justification for this is the radio access, then there is only point in choosing excerpts from those headers that may go over the radio access. In 3GPP only some of these headers do, others are removed at the P-CSCF. For example, the Visited network identifier and the charging headers never go over the radio access.

Conversely, if the hint is the P-Preferred-Identity, then that needs to be included in this group.

Keith

> -----Original Message-----
> From: Carsten Bormann [mailto:cabo@tzi.org]
> Sent: 16 July 2002 03:39
> To: rohc@ietf.org
> Cc: sipping@ietf.org
> Subject: [Sipping] P-Headers in the static SIP/SDP compression
> dictionary
> 
> 
> In preparing draft-ietf-sipping-sigcomp-sip-dictionary-03.txt, there
> has been some discussion about the best way to handle P-headers.
> Obviously, P-headers are private, so maybe they should not be part of
> a static SIP dictionary that all SIP implementations that support
> compressed signaling have to use.
> 
> On the other hand, P-headers are likely to be present in many messages
> exchanged over low-bandwidth cellular links.  Not supporting them in
> the static dictionary increases the size of these messages.
> 
> So what should we do for -04?  Here is a proposal for a compromise
> position.  Instead of bloating the dictionary by including every
> P-header in sight, we could collect a set of strings likely to turn up
> in P-headers.  Here is my provisional list:
> 
> [CRLF]P-
> 
> Access-
> Associated-
> Called-
> Charging-
> Function-
> Party-
> Visited-
> Network-
> 
> Adresses:[SP]
> URI:[SP]
> ID:[SP]
> Info:[SP]
> Vector:[SP]
> 
> This means that your typical P-header will need 3 to 4 dictionary
> references instead of one if we just pasted all of them in.  This may
> (depending on the compression algorithm) still be a sizable win.
> The dictionary size will only increase moderately, in particular since
> some of these strings are already present in the dictionary.
> 
> Comments?
> 
> WG meetings where this could turn up:
> ROHC, Wed 0900-1130, Room 502
> SIP, Wed 1300-1500, Room 301/302
> 
> Gruesse, Carsten
> 
> 
> 
> _______________________________________________
> Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
> This list is for NEW development of the application of SIP
> Use sip-implementors@cs.columbia.edu for questions on current sip
> Use sip@ietf.org for new developments of core SIP
> 

_______________________________________________
Rohc mailing list
Rohc@ietf.org
https://www1.ietf.org/mailman/listinfo/rohc