I had some hopes of RDF/A but stopped following its progress several months ago as it didn't seem to provide an easy way to really bridge the gap between HTML and RDF. My use case was (and is) to be able to markup html in a way which allows me to automatically (and without too much effort) generate context menus or tool-tips à la "Show me more info about this person" etc. I'm not sure if Ian's proposal provides a solution for this, but at least it seems to be much easier to grok than RDF/A, and not requiring XHTML2 is IMHO a huge plus.
It would perhaps have been helpful (or at least polite) to coordinate the effort with the RDF in XHTML Taskforce a bit. The W3C is working on an own approach for quite some time now (which may as well be the reason for Ian to go it alone..). However, the nice thing about GRDDLy approaches is that you don't really need a large user community for each format. For RDFers, it's another long-tail example, as long as the transformation process can be automated. Every triple counts ;)
<span id="p1">
<a rel="foaf-weblog" href="http://jd.com/blog" class="foaf-name">
John Doe
</a>
</span>
seems to be enough to implement the use case I mentioned. That's really cool as I'll now finally be able to hook an RDF store to my HTML publishing tools. It'll need some behaviour-like extensions, but it should be doable now without too many problems!My publishing system uses a homegrown markup language internally which is sort of a trade-off between simplicity and flexibility, e.g. a Web link is generated by
[[http://www.example.com/ Link label goes here]]
The HTML for a blog-identified person as illustrated above could then perhaps be created from something as simple as
[blog http://jd.com/blog John Doe] said ...Hm, generating the internal markup could be supported by some SPARQL-based "suggest persons as you type" feature, but that's of course a completely different story (and to be told another time).