RE: [office] DSIG proposal - URIs, Packages, and Namespaces - Proposal

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

I feel really thick-headed.  I wandered all over the place because of the
"existing implementations" concern, when the answer is right in front of us.
Here is my understanding of the appropriate resolution:

1. The introduction of the ODF usage of xml-dsig will be via the proposed
namespace and the xml-dsig namespace, as currently specified in the schemas
for the DSIG proposal:

   1.1 The ODF 1.2 namespace for ODF DSIG will be
"urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0".

   1.2 The xml-dsig namespace is "http://www.w3.org/2000/09/xmldsig#"; 

   1.3 For example, the element introduced by the tag

   <DSIG:document-signatures 
 
xmlns:DSIG="urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0"
       xmlns="http://www.w3.org/2000/09/xmldsig#";>

       introduces a DSIG in accordance with the ODF 1.2 specification.

2. The interpretation of URIs within a <DSIG:document-signatures> in an ODF
package subfile will be as prescribed in section 2.6 of draft 6 of the ODF
1.2 Package specification.  (This is equivalent to section 17.5 of earlier
specifications, and clarification in response to the SC34 defect report will
not impact DSIG.)  There might be additional restrictions on the forms of
URI that are usable, but the restriction must not conflict with the rules
for interpretation of URIs and the additional rules already incorporated in
section 2.6.  That is, the URI value can be used correctly without knowledge
of the restrictions on providing such a value.

3. IN THE EVENT that usage of namespace
"urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0" already exists
in documents produced by existing software, the ODF 1.2 specification will
choose a different namespace (e.g., 
"urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0") and maybe,
just to be safe, we should change it anyhow.

That's it.  That's how we avoid any confusion with an existing
implementation that uses a private namespace (as in OO.o 2.4) or one that
has already employed the originally-proposed ODF DSIG namespace (in 1.1,
above).  By making sure that an existing implementation will not be using
the approved ODF DSIG namespace, there is room to recognize the difference
in implementation and for the product-specific legacy forms to be
distinguished and dealt with appropriately by products that are implemented
to do so.

 - Dennis

PS: Duhhh.


-----Original Message-----
From: Dennis E. Hamilton [mailto:dennis.hamilton@acm.org] 
http://lists.oasis-open.org/archives/office/200901/msg00050.html
Sent: Wednesday, January 07, 2009 13:35
To: 'office TC'
Cc: 'Jomar Silva'; 'Bob Jolliffe'; Michael Brauer
Subject: RE: [office] DSIG proposal - FYI - Root Element - IMPORTANT!

Woops, that's an OpenOffice.org-specific namespace that you mention as being
in the package specification!  If that's the legacy implementation, I think
we are excused out of the box with 

<document-signatures
xmlns="urn:oasis:names:tc:opendocument:xmlns:digitalsignature:1.0"> 

(corrected to use :office: or a resolvable URL if required by OASIS) as used
in the current Package specification draft section 2.4 and in the dsig
schema file.  I was concerned that there were implementations using an
ODF-specified namespace already.  I should have paid more attention to the
URI in your remark about <document-signatures>.

I think we should go with what 17.5 has said and what Package Draft section
2.6 does say about IRIs in files of the package.  (The manifest:full-path
attribute is not of type IRI, anyURI or anything like that, so that is its
escape clause, an appropriate one for the special case of the manifest.)

 - Dennis

PS: Even if the OpenOffice.org-specific namespace and dsig files are being
used in OpenOffice.org 3.0, they are implementation-specific and not covered
by the ODF 1.2 specification or any use of office:version="1.2", I say.
There is a safe way to differentiate that implementation from an ODF 1.2 one
in some future OO.o release.  I am assuming that the current implementations
do not use the ODF namespace.  Am I mistaken?  (I also think that the
adjustment between and even dual support of the two implementations is not
that difficult, but I really can't speak for the OO.o team.  Please tell me
that the OASIS namespace for <document-signatures> isn't already being used
in OO.o documents.)


[ ... ]