- From: Richard D. Brown <rdbrown@Globeset.com>
- Date: Mon, 23 Aug 1999 10:57:54 -0500
- To: "'Richard Himes'" <rhimes@nmcourt.fed.us>
- Cc: <w3c-ietf-xmldsig@w3.org>
Rich, I do not think that we should address this problem at the xmldsig level and would rather let the upper layer defines an adequate naming-scheme. This is the position adopted by IOTP. However, this implies that all the elements being combined have been originated by a same application (true in most circumstances). When the elements are from diverse origins, I tend to feel that the application consists of an 'interchange' (xMail :-) and, in such circumstances, would promote encoding and packaging. Sincerely, Richard D. Brown > -----Original Message----- > From: w3c-ietf-xmldsig-request@w3.org > [mailto:w3c-ietf-xmldsig-request@w3.org]On Behalf Of Richard Himes > Sent: Thursday, August 19, 1999 12:27 PM > To: w3c-ietf-xmldsig@w3.org > Subject: ID Mapping > > > I posted this problem recently and didn't get a response, so I'm > guessing it hasn't been addressed. > > Suppose that there are two (or more) signed elements in two (or more) > different documents, that they are signed by local reference (href to > #id), and that they are to be combined to form a new > document, perhaps a > legal document that has collected a history of signed events for a > brief. Suppose further that it is inescapable that, in general, > duplicate ids will result, which is not well formed XML (and confuses > the signature algorithms.) These ids will have to be > renumbered, which > will break some of the signatures. > > I believe we should include a mapping element for each affected > <Signature> element, which is outside the manifest (not > signed), such as > > <IdMap>4=1 5=2 6=3</IdMap> > > Thus, if this signature signed elements with id='1', id='2', > id='3' in > the original document, and these ids had to be changed to 4, 5, and 6 > respectively in the combined document, the map would allow the ids (of > these pieces) to be converted back to their original state for > authentication. AFAIK, the map would be unable to "lie" (breach > security) and still obtain valid signatures for the > referenced elements. > > Thanks, > Rich >
Received on Monday, 23 August 1999 11:58:55 UTC