- From: Christopher B Ferris <chrisfer@us.ibm.com>
- Date: Fri, 20 Dec 2002 08:22:07 -0500
- To: "Cutler, Roger (RogerCutler)" <RogerCutler@ChevronTexaco.com>
- Cc: www-ws-arch@w3.org
- Message-ID: <OFB9477C5C.063818D8-ON85256C95.00463743-85256C95.004948B4@rchland.ibm.com>
Hmmm... I'm no P3P expert either. However, my understanding is that a P3P expression asserts policy of a site, or subsection of a site, not a client/user. If B's policy says something like: "won't give your phone number to 3rd parties" then clearly B would be violating its stated policy if it were passing this information on to another service such as C, regardless of what that third party service's P3P policy might claim and regardless of who C is. Now, if B's policy stated: "we do not use your email adresss to send you obnoxious spam on a daily or more frequent basis" and did NOT make the claim that: "we won't send your email to third parties" and B then sent A's email address on to C to fulfil the request (or for any other reason AFAIK) and C's policy did not make such a claim, then it isn't clear to me that B is violating its stated policy by giving the information to a third party who's policy was inconsistent with its own. Of course, the P3P REC does not provide a technical solution to automate enforcement of a policy, as Roger has correctly cited, nor does P3P cover transferring data in its scope, so this aspect of the discussion (transitive nature of P3P) may all be moot. Cheers, Christopher Ferris Architect, Emerging e-business Industry Architecture email: chrisfer@us.ibm.com phone: +1 508 234 3624 "Cutler, Roger (RogerCutler)" <RogerCutler@ChevronTexaco.com> Sent by: www-ws-arch-request@w3.org 12/19/2002 06:47 PM To www-ws-arch@w3.org cc Subject P3P Hairball?? I can't figure out if I know a little more about P3P than some of the people on the call today -- or a lot less. I certainly am not a P3P expert, but I have looked at how it works. And it impressed me how little of it is truly automated. It seems to me that there is considerable possiblity for P3P to mix ungracefully with a machine-to-machine automated web services environment. For example, if A invokes a web service at B, sending some information to B and expecting some information back -- B may, under the covers, call a web service at C. Although I have not seen it explicitly called out (and maybe it should be), I think that this may be truly under the covers. That is, I don't think it is reasonable to force B to tell A that it has called, or is going to call, C. I think that such a requirement could cause a lot of trouble for commercial applications, including security concerns. In that case, how in the heck does the P3P policy of C get into the act? My understanding is that this is not at all trivial -- and maybe even beyond the scope of P3P as it stands. As usual, my apologies if I have flawed understanding of what's going on here and am just spreading confusion.
Received on Friday, 20 December 2002 10:34:06 UTC