RE: [Sip] Privacy statements and History (draft-ietf-sip-history- info-04.txt)

Roland and Sebastien,

I too agree with the proposal to change the strength of the referenced
statement from section 4.3.3.1.1  from MUST to SHOULD. This change is
consistent with the other normative statements in section 4.3.3.1.1, which
are SHOULDs rather than MUSTs.   I'll make that change in the next rev
unless someone raises concerns over that change. 

On the second concern, I'm not clear what the proposed change is for the
following:
[SG] Another issue is the decision to add hi-entries in a privacy context. A
proxy changing the target (i.e. the proxy is responsible of the resource
reflected in the received Request-URI) of a request which contained a
"privacy=history" header MAY add a history-entries provided that it knows it
can rely on other entities within the trust domain to apply the requested
privacy. This affect item 4 in the list on conditions of section 4.3.3 and
4.3.3.1. 

Item 4 in section 4.3.3 describes the general situation, thus the strength
(MAY) describes the general fact that the use of privacy is optional.  The
normative text provided in 4.3.3.1, provides the general model for adding
hi-entries, with the privacy specific considerations for adding the entries
described in 4.3.3.1.1.  I think what you're suggesting is changing the
strength of the following statement in section 4.3.3.1:
" The hi-entry MUST be added following any hi-entry received in the request
being forwarded."
I can see that in the context of privacy considerations, this might seem
misleading.  That MUST was originally a SHOULD but was changed (as discussed
on the list and at IETF-60) to MUST based on issue JRE- 4, so a simple
change of MUST to MAY would not at all work.  I thought it was clear from
the statement at the beginning of 4.3.3.1, that this section is describing
the scenario under which the general privacy considerations had been
evaluated and the intention was to add an hi-entry, thus the statements
should be read in the context that the general screening indicates that an
hi-entry SHOULD be added and if one is added it MUST be added following any
entry that's already in the request to preserve the ordering.   If you have
explicit changes to the text that you think would further clarify the
functionality, we can discuss. 

Roland, if you have specific text on how I could incorporate a reference to
RFC3323, we can consider that; fundamentally any discussion of privacy
depends on RFC 3323, so it's not clear to me how you were suggesting we
incorporate such a reference.  The text in this document (history-info)
should accurately describe the processing impact of the new "history"
priv-value. 

Thanks for your input,
Mary 

-----Original Message-----
From: Jesske, R [mailto:R.Jesske@t-com.net] 
Sent: Wednesday, November 10, 2004 7:59 AM
To: sebastien.garcin@francetelecom.com; Barnes, Mary [NGC:B601:EXCH];
sip@ietf.org
Cc: VL-T-Com-T-TE332@vli.telekom.de
Subject: AW: [Sip] Privacy statements and History
(draft-ietf-sip-history-info-04.txt)

Dear Mary and Sebastien,
here are my Comments to Sebastien's statements:

Your first proposal I can accept from my point of view.
On you second issue my impression is that this issue must be seen with
regard to RFC3323 where privacy and the trust concept is described. Perhaps
a reference to this document should be included.
On your last point, I think we can also refer to RFC 3323.

What are you thinking?

Best Regards

Roland

 ---- Mary snipped forwarding info to keep message smaller-----------  

-----Original Message-----
From: GARCIN Sebastien RD-CORE-ISS
[mailto:sebastien.garcin@francetelecom.com] 
Sent: Monday, November 08, 2004 10:11 AM
To: Barnes, Mary [NGC:B601:EXCH]; sip@ietf.org
Subject: [Sip] Privacy statements and History
(draft-ietf-sip-history-info-04.txt)

Hi mary, all

When reading draft-ietf-sip-history-info-04.txt, I have trouble in
understanding some of the statements which relate to the forwarding rules
for history-entries subject to privacy. It is an important requirement that
History-entrie(s) with a Privacy=history, session, or header are indeed
forwarded to entities which belong to the same trust domain. The removal of
specific history-entries should only occur if the peer does not belong to
the trust domain.

In the current text (section 4.3.3.1.1) :
If a request is  being forwarded to a Request URI associated with a domain
for which the proxy is not responsible and there is a Privacy header in the
request with a priv-value of "session", "header" or "history", the proxy
MUST remove any hi-entry(s) prior to forwarding. 

The current wording is misleading since it gives the impression (maybe
intentionnal) that it is not possible to forward history-entries with
Privacy statements to domains under the responsability of e.g. another
operator belonging to the same trust domain.  

[MB]: Current wording is consistent with terminology in RFC 3261 in terms of
describing who is able to change the Request URI in a specific request
(based on section 16.5): 
   " A proxy MUST NOT add additional targets to the target set if the
   Request-URI of the original request does not indicate a resource this
   proxy is responsible for.

      A proxy can only change the Request-URI of a request during
      forwarding if it is responsible for that URI. "
Since History-Info (and associated privacy) are only added to the request,
when an entity that is allowed to change the Request-URI retargets the
request, it seemed sensible to use consistent wording to explain that.   
[/MB] 
 [SG] I have no problem with the statements above. My concern is that if
there is a hi-entry already embedded in the request with a Privacy
statement, then, it should be up to local policy to decide whether or not a
proxy shall pass on those hi-entries to a trusted domain. The sentence in
section 4.3.3.1.1 precludes this. I would propose to lighten the strenght of
the sentence as follows:

If a request is  being forwarded to a Request URI associated with a domain
for which the proxy is not responsible and there is a Privacy header in the
request with a priv-value of "session", "header" or "history", the proxy MAY
remove any hi-entry(s) prior to forwarding. 

[SG] Another issue is the decision to add hi-entries in a privacy context. A
proxy changing the target (i.e. the proxy is responsible of the resource
reflected in the received Request-URI) of a request which contained a
"privacy=history" header MAY add a history-entries provided that it knows it
can rely on other entities within the trust domain to apply the requested
privacy. This affect item 4 in the list on conditions of section 4.3.3 and
4.3.3.1. 

The concept of "trust domain" should be used when discussing the forwarding
rules pertaining to information subject to privacy. Furthermore, the
requirement for forwarding history-entries to trusted entities should be
stated more clearly in the draft. 

[MB]: The whole concept of what defines privacy in terms of the proxy's use
of the privacy header is outside the scope of History-Info functionality and
really a matter of local policy.   I think the functionality that you want
is a matter of local implementation and policy in terms of operators
establishing this "trust domain" model to which you refer.  History-Info
defines the mechanism to ensure the privacy of the requests, but it doesn't
explicitly define how the proxy knows whether it is responsible for that
resource.    I don't think this is a matter of standardization.  I thought
the use of the term "domain" rather than "resouce" would be helpful, but
perhaps changing it to the more general "resource" would resolve this
concern and/or a statement clarifying what I've just described should be
added in the draft. 
[/MB]
[SG]   Changing the term "domain" to "resource" does not solve the problem I
mentionned above. In order to reflect current operator requirements, the
draft should not PRECLUDE the fowarding of existing history information
towards trusted domains.

Thank you for clarifying this point.

Best regards,
sébastien

------Remainder of non-related part of this thread has been deleted by
Mary------------------