RE: digest requirement

On September 18, 2001 12:06, Joseph Reagle wrote:
> On Sunday 16 September 2001 11:23, Amir Herzberg wrote:
> > I think Joe's scenario would work. Few comments:
> >
> > 1. Don't we need to copy the ID's to the EncryptedData tags for the
> > references to work, e.g.:
> >
> > <AlphabetiSphagetti>
> >   <A id="a"/>
> >   <EncryptedData id="b" xmlns='http://www.w3.org/2001/04/xmlenc#'
> >    Type='http://www.w3.org/2001/04/xmlenc#Element'>
> >     <CipherData>
> >       ....
> 
> For the references to work to what end? For the Signature 
> Transform [1], or something else?
> [1] http://www.w3.org/Encryption/2001/Drafts/xmlenc-decrypt

No, simply for the reference you've put in the manifest to be valid,
e.g.:
<Reference URI="foo.xml#b"> 

Without putting the `id="b"` in the EncryptedData I think this reference
won't identify this element. 
> 
> > 2. What if we want the signature to also include a regular
(mandatory,
> > not Manifest) SignedInfo for parts of the document which are never
> > encrypted? <skip>
> Are you familiar with [1]? 
> [1] http://www.w3.org/Encryption/2001/Drafts/xmlenc-decrypt

Familiar, but forgot about it when I wrote, sorry... Sure, this ([1])
can be used to sign the document with non-encrypted parts and use the
ciphertext for the encrypted parts. However, this does not allow to
change the encryption (e.g. to another key) or remove it, which was one
of my goals (and a pretty natural one). For this I still think that we
must either use DigestValue in the EncryptedData, and a transform to
sign only the DigestValue, or a transform to remove the entire
EncryptedData and sign it only via Manifest as you suggested. 

Best, Amir Herzberg

Received on Monday, 24 September 2001 06:35:00 UTC