Isn't the canonicalized output only used for digesting? And a SHA1 digest size is always 20 bytes regardless of the input size. So what do we benefit by reducing the XML size? 

In the streaming proposal I have suggested that we combine the canonicalization and the digesting for efficiency, so as the canonicalizer emits bytes for each element, those bytes are immediately fed to a digestor which maintains a running digest, and at end you end up with a digest for that reference, without needing to allocate temporary memory to store the entire canonicalization output.

Maybe the comment was about using canonicalization in a different context, i.e. not for signing/verification.


Thomas Roessler wrote:
On 2008-08-27 10:44:08 -0400, Frederick Hirsch wrote:

Isn't the answer here that XML needs to remained well-formed, so no?

If the output of Canonicalization needs to be well-formed XML, then
the answer to the comment is indeed "no".  If the requirement is
relaxed, then the answer is probably "maybe".

regards, Frederick

Frederick Hirsch

On Aug 15, 2008, at 7:36 AM, ext Thomas Roessler wrote:

fyi, this came in as a comment on Canonical XML.

FWIW, I'll make sure that public-xmlsec-comments gets subscribed to
the comment mailing lists for all the specs that we are chartered to
deal with, to make it easier to follow them all.

-- Thomas Roessler, W3C  <>

----- Forwarded message from Amol Patil <> -----

From: Amol Patil <>
Date: Thu, 14 Aug 2008 20:14:06 +0100
Subject: xml canonicalization - proposition - elimination of element 
name from end tag
List-Id: <>
Archived-At: <>
X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.1.6

Can xml canonicalization further be extended to eliminate the need of 
name of the element in end tag / make it optional?

Objective here is to reduce the size of overall xml content


There is a slight loss of readability here at the gain of reduced  


----- End forwarded message -----