W3C home > Mailing lists > Public > w3c-ietf-xmldsig@w3.org > July to September 1999


From: Peter Lipp <Peter.Lipp@iaik.at>
Date: Wed, 25 Aug 1999 18:16:48 +0200
To: "W3c-Ietf-Xmldsig@W3. Org" <w3c-ietf-xmldsig@w3.org>
Message-ID: <001d01beef15$3b9ba3c0$87981b81@iaik.at>

sorry for being so quiet recently. I was quite busy and now took some time
to study the pile of material. I am still on the learning curve with regards
to XML-Pointers or Canonicalizers (so sorry about any blunder i made), but I
still want to push it off to you...

Requirements: I would consider it an essential requirement for any
canonicalizer to be absolutely loss-free. XML-C14n states in the
introduction: "with potentially some information loss". This is absolutely
unacceptable for digital signatures, and while many have expressed their
reserved feelings towards canonicalization (especially being mandatory),
this should be put into the requirements doc - and also later into the
security requirements for canonicalizing resources.

Ed Simon mentioned on the list a while a go:
	> If the legal value of a signed XML document
	>is dependent upon how it was presented through
	>a stylesheet, that stylesheet MUST be signed too.
but C14N removes such stylesheet information according to 5.3

> I would expect that in many applications, the validity and usefulness of a
> digital signature from a LEGAL perspective will require capturing not only
> the content of the XML source but also its presentation to the user

European Guidelines talk about signature environments should present the
user the information he is going to sign (what you see is what you sign).
Now considering the potential implications of pulling in external references
(section 4 of c14n) ...... is it likely that an application might do that
when displaying some XML? Maybe, I have no idea, but it makes me nervous in
case it shouldn't and the signature module than pulls in (lots of) documents
by indirection -  I assume that in case e1.xml in c14n/section 4 could again
contain entity references?

I must admit that canonicalizers makes me very nervous generally .....
except the NULL-variant :-)

I thought I had understood, but the more I think about it  the less I do:
what do we need that for in the first place? To save some space for some
apps that tend to manipulate and store these things in different forms (DOM)
only? Couldn't they simply save the original data somewhere?

Dr. Peter Lipp
Email Peter.Lipp@iaik.at
Phone +43 316 873 5513
Fax   +43 316 873 5510
Web   http://jcewww.iaik.tu-graz.ac.at
Received on Thursday, 26 August 1999 02:25:17 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:09:55 UTC