- From: Ari Kermaier <arik@phaos.com>
- Date: Thu, 30 May 2002 13:23:49 -0400
- To: reagle@w3.org, merlin <merlin@baltimore.ie>
- Cc: "Takeshi Imamura" <IMAMU@jp.ibm.com>, xml-encryption@w3.org
Joseph, I implemented the decryption transform using the following interpretations of the spec: 1) XPointer support that is OPTIONAL in Signature remains OPTIONAL in Decryption. The Phaos implementation does not currently support the OPTIONAL XPointer syntax, but I saw no reason to preclude other implementations from supporting it. 2) The description of the process of selecting e in X but not in R implies that the iteration continues until no EncryptedData element satisfying the condition can be selected. I.e., the contents of super-encrypted elements are automatically decrypted (in an arbitrary order relative to other top-level EncryptedData elements). I had not considered the sort of document structure that Merlin used to illustrate potential ambiguities arising from the processing rules. Merlin makes a good point on this subject -- I'm just not sure I understand his proposal for dealing with super-encrypted elements by using multiple DecryptTransform elements in the dsig:Reference. Ari At 12:13 PM 5/30/02 -0400, Joseph Reagle wrote: >On Wednesday 29 May 2002 07:36 pm, merlin wrote: > > However, all these changes boil down to a large refactoring > > of the decryption transform spec; and, if I am unique in > > my viewpoint, then I am at an impasse and must cede to the > > majority view and stand by my former stated views on the > > subject. > >I think this is a great discussion; it's what happens when "implementors >hit the spec." <smile/> Beyond your new requirement which I'll address in a >separate email, I'm generally happy with the direction of what you propose, >but I'm most interested in the thoughts of other implementors, which at >this point [1] include IBM and Phaos > >[1] http://www.w3.org/Encryption/2002/02-xenc-interop.html > >1. In the present WD, we say we support the xmldsig profile XPtr (which is >same-document XPointers '#xpointer(/)' and '#xpointer(id('ID'))'. Anything >else is optional. You seem to be using more sophisticate pointers, are you >advocating a change in this level of support? (BTW: I think I've heard that >folks are reviving the XPtr work by breaking it into different >profiles/specs.) I'd prefer to keep this real simple and rely upon IDs. >2. In decryptIncludedNodes() why did you switch the order of steps 2 and 3? >(Makes it difficult to see the diffs, but maybe you had a reason.) >3. In decryptXML, if I have three elements I'm decrypted, via process one >do they all go into one octet stream, or is a group of 3 octets sets? > >-- > >Joseph Reagle Jr. http://www.w3.org/People/Reagle/ >W3C Policy Analyst mailto:reagle@w3.org >IETF/W3C XML-Signature Co-Chair http://www.w3.org/Signature/ >W3C XML Encryption Chair http://www.w3.org/Encryption/2001/
Received on Thursday, 30 May 2002 13:22:11 UTC