- From: Amir Herzberg <AMIR@newgenpay.com>
- Date: Sun, 20 May 2001 09:21:40 +0300
- To: "'Joseph M. Reagle Jr.'" <reagle@w3.org>
- Cc: "Xml Encrypt (E-mail)" <xml-encryption@w3.org>
Joe said: > From: Joseph M. Reagle Jr. [mailto:reagle@w3.org] > At 09:48 5/15/2001 +0300, Amir Herzberg wrote: > I was trying to focus on the added features of the proposal, > so when you say: > > >-- A recipient with the decryption key can validate that the entire message > >(including the encrypted part) was signed > > When used with XML Signature? (When you say encrypted part, do you mean the > plain or cipher data?) When using XML DSIG; and validation is for the plaintext. > > >-- A recipient without the decryption key can only validate the > >non-encrypted parts of the message. > > Well, he can validate the whole document, which is the version with > EncryptedData included. Of course. I meant, that this recipient is not able to validate (or see) the plaintext. <skip> > There's really two parts of this proposal which I'd like to break apart: > 1. Integrity: hashOfRandomized (let's call it DigestMethod and DigestValue > in CipherData) > 2. Morphing Feature: changing Encryption information without > breaking the signature. Agreed. But there's a third feature, which you may consider just a variant on Morphing: 2.1 Trimming: removing (most of the) Encryption information without breaking the signature, creating a version of the message which does not contain the encrypted information. > > The first part, I think makes sense and is fairly straightforward. I can add, that for the applications I have in mind currently, only Integrity, and to lesser extent Trimming, are applicable. > > The morphing necessitates a transform, but just to start with natural > language: > (1. Resolve the Reference URI). > 2. Find any EncryptedData children with a HashOfRandomized child. > 3. Replace the EncryptedData element with its HashofRandomized child. > > My tenative concerns with this approach: > 1. How would this interact with the encrypt-sign transform, any side-effects > or does it become unwieldy and complex when you consider both? Sorry, I lost reference to encrypt-sign transform, please re-send it so I can review this. > 2. Performance? This proposal should allow removal of encrypted data for performance reasons. The transform should not be too heavy, and anyway is optional, of course. > 3. It's clear we need a more complete specification. For the tranform, yes, but I think this can and should be done after we agree on allowing the HashOfRandomized (or better name) element. As you said, this tag can be used also for achieving just the Integrity goal; for that the transform is not needed. Best regards, Amir Herzberg CTO, NewGenPay Inc. See demo and lectures/overviews/tutorials on crypto-security for mobile, e-commerce, etc. in http://www.newgenpay.com/mpay/course/course.html
Received on Sunday, 20 May 2001 02:18:10 UTC