Re: [office-metadata] Atom and document/feed IRIs

[Michael Brauer - Sun Germany - ham02 - Hamburg <Michael.Brauer@Sun.COM>]

Maybe the following is helpful in our discussion (at least it clarifies 
my personal position): Our URU/IRI handling currently is based on 
RFC3986 (http://www.ietf.org/rfc/rfc3986.txt), which in section 5.1 
defines how a base IRI is established.

The text is actually to long to be repeated here, but one conclusion we 
can draw from this is that all ODF documents have a base URI, that's the 
retrieval IRI (level 5.1.3), and that we may specify a base IRI in the 
content (level 5.1.1) that then overwrites the retrieval IRI.

 From my understanding, the open issue actually is whether we want to 
make it mandatory that all documents specify a base IRI in the content. 
If we do so, then we have to take into account that ODF document may 
already contain relative IRIs, and they are actually used. If we specify 
a base IRI in the content and if we want to stay consistent RFC3986, 
then we have to apply the base IRI to all relative links in the 
document, and not to the relative links for metadata only. This means 
that relative links to for instance images do not work any longer if we 
add base IRIs that are just identifiers for the purpose to get unique 
subject IRIs. If make that base IRI mandatory, then we actually loose 
the possibility to use relative IRIs in document for other purposes than 
metadata at all. That's something I think we actually should avoid.

If we find a way to specify a base IRI that is used for metadata only 
without getting inconsistent to RFC3986, then I would have no objections 
for doing so from the technical perspective, although I have some 
difficulties to imagine how we can make sure that these IRIs remain 
unique and meaningful. To give an example: Let's assume I have a 
document at location "A" with the metadata IRI "a". I edit the document 
and save it again. I assume this new revision of the document keeps the 
IRI "a". I now copy the document to a location "B". The metadata IRI 
remains "a". I make some other changes to the document and save it 
again. Since I cannot differ the original document from my copy the IRI 
remains "a", but since I now have two different documents, the IRI is 
not unique any longer. One may think of other rules now, but I fear 
whatever rules office applications use here, they run into similar 
problems. For this reason, I would feel much more comfortable if we 
allow an option base IRI, but don't make it mandatory.

Michael



Svante Schubert wrote:
> 
> Patrick Durusau wrote:
>> I thought we were talking about assigning IRIs to identify documents? 
>> Rather than path statements.
> You might be right. I took this for an argument for IRI against relative 
> links to xml:id on ODF elements.
> 
> Isn't the identification (IRI) of mails and feeds given after they are 
> modified, when they are read-only in their remaining life-cycle?
> Regarding unique identification of ODF documents, generating hash value 
> appear more consistent instead of IRIs.
> Would it not make more sense for us to use an IRI given from an RDF 
> application, like a Invoice IRI?
> I say, leave the identification to the RDF application, which can 
> clarify what makes an ODF document unique in their eyes.
> 
> Svante.
> 
> 


-- 
Michael Brauer, Technical Architect Software Engineering
StarOffice/OpenOffice.org
Sun Microsystems GmbH             Nagelsweg 55
D-20097 Hamburg, Germany          michael.brauer@sun.com
http://sun.com/staroffice         +49 40 23646 500
http://blogs.sun.com/GullFOSS