Re: Dropping RDF mapping from microdata spec

On 28 July 2011 19:37, Jeni Tennison <> wrote:
> Nathan,
> On 28 Jul 2011, at 17:16, Nathan wrote:
>> I want to create a scheme for microdata authors to use, I want it to be machine readable and dereferencable, where's the spec I should follow?
> There's no spec for that (so far as I know): if you create a vocabulary, you write a document that explains how the vocabulary works, maybe supply a preview/validation tool like the Google Rich Snippets validation tool if you need to.

That's pretty much in line with the microformats approach (which may
be influential in this context):

> There's also (to explain the thinking behind the way microdata works here) no need for automated discovery of such a schema. A consuming application either has hard-wired understanding of the vocabularies that it finds in a page or not. If it doesn't, it must not follow any links (eg the item type) to get any more information in order to understand the vocabulary. I think the rationale there is that if they could, it might lead to a dependence of the behaviour of the application on network connectivity.

Ew. Download & cache surely preferable.

> Any code that consumes the vocabulary will naturally validate (and interpret, particularly to map to an appropriate data type) to whatever degree is useful.

Makes sense, but I suspect any mention of validation will make the
HTML folks run a mile -  maybe draw back to just sanity checking :)



Received on Friday, 29 July 2011 09:13:00 UTC