Reported by Dorothy Gellert

SIPPING Session 2, IETF 64

Wednesday, November 09, 2005

Updade to Agenda:  “Implementing  prack” will be presented at the end of session from Rohan.

Elwell-sipping-service-retargetting-00:  Chairs presenting these slides for John Elwell

Long thread in sipping WG mailing list

Authors realized work could be met by voicemail-uri draft

D.Wilis:  Output draft will  be AD sponsored informational?  

Gonzales:  Yes.

Keith from Lucent:  Why should this be AD sponspored?  TISPAN wants this

Rohan:  Reminder, we agreed not to take this on as a W.I.  

Dean:  Futher Review Humm.  Result:  Send to AD and IESG.  

Cullen:  We shouldn’t be expanding the scope.   No new technical content.

Extending-the SIP Reason heade with Warning coder – Jani Hautakorpi

Cullen:  Controller to refer to 1^st guys to talk to 2^nd guy if there is a problem?

Gonzales:  Use case is RFC 3326.  For completeness, we can provide state information in 2 places.  Warning codes are useful.  

JR:  This is contradictory, since not many people use warining codes.

Gonzales:  Wants to know if more people want to use warning codes;  Wants an explicit discussion on this.  If we want this, how do we handle warning codes?

Miguel:  TIPSAN & OMA POC looking for specific codes. 

JR:  We have response codes.  1) to tell something happened to a person  2) tell automata what happened?

Sean:  We are running out of codes and we need more information than the reason code can give us.

JR.  Not much space in IANA registry.  Application specific response codes that other SDOs don’t want to bring to IETF…   some disagreement about this opinion from the group.  

Rohan:  IANA doesn’t reflect new codes until they are far into the IESG process  Will someone look at  $XX and 5XX namespace?    Robert Sparks agreed to do this

GRUU Reg Event Package  - Paul 

Chnages since Fillied in IANA registration.  JR says IANA registration is wrong.

Issue on list:  Reqest to add report identity.  Paul thinks this is out of scope, since registrars don’t normally keep this info.  

Paul:  Stay as we are.

JR agrees.

D. Willis:  We have consensus to go with draft as is and will schedule WGLC.

File Transfer Services in the context of SIP based communication:  Markus Isomaki

User to user communctiation services, would be nice to integrate user to user services in this framework, such as file transfer.  Page mode and session related scenarios presented.

Joel:  Bothers him, if 2 partys communicate, there a lot of things they can do that are applications .  We don’t want to specifiy how to do all these things in SIP.   Out of scope as to how to invoke every application in SIP

Eric Berger:  What does this get us?  Why are we standardizing this in Sipping?

D. Willis:  

JR:  He is not saying we need something new, in IM & presence have file transfer.  Should this be in simple?

Francioise:  a lot of sip mechanisms do this, doesn’t think it requies new protocol.

Cullen:  Good to have this in any group

Hesham:  This should be informational.

Markus:  yes.

D. Willis:  new MIME type would require standardization

Is MSRP going to be used for file transfer?   

Markus:  Yes, that could be one of the ways to do this. 

JR:  If you want to post to a server what url to you use?  Ok to do a requirements document.

Markkus will start a requirements document and scenario document.  Where should he send these drafts?

Dean:  Send to Sipping WG to start.

Sip Identity Usage in Enterprise Scenarios:  Steffen Fries

End to end authenitication, identity provisioning, security parameter bootstrapping.

Cullen:  How could this possibly be in a corporate directory?  Can’t do this with the certificate server.  Cullen disagrees with the motivation to use the certificate server.   Not deployable

Steffan:  Not the same as a directory server

Cullen:  European rules won’t let you deploy this:

John:   User based certificatie have had difficulties deploying, this is device certificate..

Cullen:  How is a device certificate different than user based?  What is a device cert?

John:  Many users on the same cell phone can use certs to identify themselves on same hardware.

Franciose:  If you don’t want to use user certificates, you can use device certificates, don’t confuse this draft with user certificates.

Cullen:  Define device certificate.

Cullen: This is already in RFC 3261

Sean:  What is Enterprise specific in this scenario?  Don’t have to be tied to a enterprise directory…

Steffan:  Yes, you need to map identify to certificate 

Dean:  Recap:  this draft makes a temp binding between device cert and identity.

Sean:  Corporate directories do that today.  It works today.  

Roahn:  Its works but it is expensive!

Franciose:  This draft doesn’t define anything new.  

John: ( couldn’t follow argument for the draft…)

Dean:  Consensus:  An interesting topic that needs to be defined so Rohan can understand it!

Sipping-calling –party-category-01 - Rocky Wang

James Polk:  IF CBC is so sensitve why is it in the header?

Rocky:  

JR:  In any SIP network there is no way a called party can know if a hdr was inserted by a proxy or anyone else…  Use SAML to validate that a caller is a fireman, prisoner, etc….

Dean:  Have we thought about doing this with Security Solution Markup language…. SAML?

Martin Dolly:  In draft there are values that are OLI values…   Orignal Line Identification

JR:  Can we use the same parameter if it’s a CPC or OLI?

Dolly:   At the PSTN gateway you get CPC and OLI values in North America.  Some parameters will only be in OLI not in CPC.  In CPC some of these values don’t exist.   Parameter in ISUP needs to get intoCPC

JR:  Hodgepog of Crap.! Do we need to to have Data Call supported in SIP?

Steve Harris:  Also same problem in UK.  

Janet:    Things are done with CPC that also need to be done in the IP world.    Don’t need to come up with new parameter that reflects problems in the PSTN in IP

Eric:  We need this kind of information, its worthwhile to do.

JR:  Objective    we need to assert roles and attribute to callers that will affect services… Some of these are found in CPC and legacy PSTN protocols…    Find a set of paratmeters in relation to SAML

Miguel:  This is an non biased solution and it’s a 

SAML and role based stuff, may need better requirements, go do RD, or solve in SAML?  Any volunteers,  Rocky?  Yes, he will follow up on this.  

Someone volunteered Hannas who is not here. 

Dean:  Rocky, talk to Hannas to see if there is common interest…

Implementations of Prack  - Rohan

MMUSIC session this morning points were brought up based on avoiding implementation that do PRACK.

Rohan volunteered to givet update in Sipping on Prack.

Implementation with Preconditions use Prack.

IMS lite uses Prack

Complexity is why most don’t implement Prack.

- UA needs to queue or delay messages for an external event.

- UA needs retransmit queue of 18x responses

Affects offer/answer state machine.

What to do now?  

JR:   There are engineering tradeoff that are worth complexities.   

Have a way to do addition messaging prior to sending response to the oringal requests - Biling

Backward compatible with RFC 2543.

Caution, if you want to do something better, which of these 2 characteristics are not important to you?

Cullen:  Need a document.

JR:  cost of implementation doesn’t justify the Business reason.  If it wasn’t useful before doesn’t mean it can’t be useful now.

Robert:  third of endpoint implementation claim 100% of Prack, but in truth, they only implement 4 cases, not the full body.

Paul Kyzivat:  related to offer/answer confusion.    Mmusic offer/answer draft is out there.  

VJ:  Make Rohan’s state dirgram an RFC!

Payment for services in SIP  - Jason Fischi

Not many people read this draft

David Schwatz??:  Want to collect enough money, but it matter what country it comes from avioid SPIT

Joel:  Hate to make solving SPIT dependant on a payment system  Interaction between user and proxy and interaction with the payment server.  Payment infrastructure is a different ketlle of fish!   We shouldn’t work on this.   

Jason: Intention to provide a full mechanism, but its an early draft and looking for comments

Joel: Negative on this, because  Payment management protocols are different for different scenarios.    Sipping is not a Commerce payment WG!

Cullen:  Should be in the Spam architecture document.  Classic payment protocols don’t work for transaction under 30 cents per transactions.

Rohan:  2^nd half of this is out of scope.   Likes the draft, feels there are applications for it.  Information libraries like listening to songs, SPIT prevention is a terrible idea… Poor nations vs Rich countries

Heshim:  deployment of ths draft, someone will have to consume the cost… it’s a problem.  

Cullen:  SPIT arguments send comments to Spam framework doucmnet authors

David Brian:  Needs to be based on the country It is going from…

???:  payment side is larger than the spam side

Jason:  Do we think we should specify SPAM?    Should we separate mechanism at this time?  Needs to work for low value transactions.

Dean:  We are hung up on what to charge foreign SPITTERs   Defer the discussion.

Jason:  We use SAML so a proxy can act on behalf of the user or the merchant.  This adds additional cost.

Rohan:  Do we want to use SAML?  There are no objections.

Robert:  How many people are up to date on SAML?   Small amount of people

Service Identifiers – Henning 

 Cullen:  Okay to use for pizza delivery, but not for 911.  He  likes previous SOS proposal better for emergency because same code path would be executed,…   

JR:  Disagrees with Cullen.  Separate code path is preferable.

Henning:  For ER services there should be test service, before the emergency 

Keith:  Agrees with JR, disagrees with Cullen.  Separate code path is better.

Henning:  IF this is interesting he want to keep in one place (WG)

???- Long term solution to get this done could be in ECRIT or even here…  

Henning:  If in sipping he would generalize this to operator services that woud be beyond ECRIT

??? - General Services could make this useful for everyone, like pizza delivery

Henning:  Doesn’t want the pizza delivery to delay the 911 call!

Robert:  SOS needs to be moved to ECRIT

A.I  to chairs:   take this to the ECRIT WG charis to find a home for this work.