OpenDocument and XMP
Alan Lillich has just joined the OASIS OpenDocument TC. Alan is an engineer at Adobe who works on their Extensible Metadata Platform (XMP), and he joined the TC to lend his expertise towards resolving the metadata discussion that has been taking place over the past few months at the TC.
To that end, Alan has posted a long explanation of Adobe’s perspective on the problem, entitled OpenDocument metadata and XMP, which is in part a response to some of the concerns I have raised. In a nutshell, we are left some difficult choices. The TC can adopt XMP as is, and get instant metadata interoperability across a range of applications and file formats.
In doing so, however, one gives up a lot of important RDF features, and must play by the XMP rules. XMP has no support for native RDF linking facilities or typing. Likewise, it has no support for either rdf:parseType=”Literal” nor rdf:parseType=”Resource.” Finally, in the current syntax, all duplicate properties must be placed in either rdf:Alt, rdf:Bag, or rdf:Seq.
OTOH, XMP has been widely deployed across a variety of commercial and open source applications for the past few years. Modifying XMP to address some of the above limitations could involve serious backward compatibility issues for Adobe and others.
For those with the technical background adequate to assessing the trade-offs here, it would be good to comment on Alan’s piece somewhere. I suspect it will frame the discussion going forward, and we’ll all be living with the outcome for a long time to come.
update: Bob DuCharme reminded me of another big concern a number of people have raised: Adobe’s XMP toolkit only supports C++, which greatly restricts its real world usability.
