- From: Bruce D'Arcus <bdarcus@gmail.com>
- Date: Thu, 11 Jun 2009 05:09:13 -0400
On Wed, Jun 10, 2009 at 6:44 PM, Jonas Sicking<jonas at sicking.cc> wrote: ... > However, as others, I'd prefer to see these things developed > elsewhere. Mostly because the group of people with expertise in > developing a better version of bibtex is not the people in this WG. Exactly; it defeats the purpose of having an extensible metadata system, whose virtue is to allow distributed vocabulary development. > I do think it's important to show that microdata is able to express > something like bibtex. And I do think that the discussion in the past > weeks have been interesting since people haven't actually been finding > problems in microdatas ability to express something like bibtex, but > rather in the exact bibtex format itself. This is a bit like triage, though. My immediate concern has been this particular use case, and I've been assuming : that the microdata proposal will be included in HTML5. In a vacuum, I think microdata is fine technically. In the context of an existing spec that covers the same use cases (RDFa), I think it's creating unnecessary and unproductive duplication. Just to go back to the use case I'm focusing on here, it puts metadata producers and consumers in an awkward position of having to likely support two different specs; means double work with no obvious benefits. This is happening JUST as RDFa is starting to be implemented by major players, and starting to build up a head of steam in terms of tools. And to put this in some context, the only reasonable technical point that Ian has made in favor of throwing out RDFa and creating a new spec is the prefix issue. But I have a really hard time seeing how prefixes is so onerous a burden as to justify the costs (to the WHATWG, and to metadata producers and consumers) of creating and maintaining a new spec. FWIW, some possibly relevant background from the OpenDocument experience: To make a long story short, ODF 1.2 will have an extensible metadata system based on RDF/XML (for in--package metadata) and a subset of RDFa.(for embedded). Getting to this solution was a long and torturous process, and the original proposal effectively forked RDFa by requiring fully unqualified URIs for names. The technical reasons were more-or-less the same as those that drove Ian to invent an entirely new spec: that in a GUI environment where users are copy-and-pasting content, dealing with prefixes was an additional burden on implementers. In addition, people don't hand author ODF files, so prefix have no authoring benefit. In the end, though, I understand the ODF TC decided to include prefixes, since implementers found the burdens largely theoretical (OpenOffice should see an initial implementation in 3.2 I understand), and because in general the group prefers to stick as closely to existing specs as reasonable. On predefined vocabularies, we thought about doing something similar informally, but decided it was out-of-scope; better initially to put a solid extensible system in place and let developers start working with it. My work on the Bibliographic Ontology was in part done with that in mind, though has the added benefit it can be repurposed for RDFa in XHTML. Bruce
Received on Thursday, 11 June 2009 02:09:13 UTC