On 15-06-22 10:16 AM, Lou Burnard wrote:
On 19/06/15 20:08, Martin Holmes wrote:
A possible example might be:
Redefine <geo> so that it explicitly contains GeoJson:
http://geojson.org/geojson-spec.html
I'm not sure whether this qualifies as a datatype, strictly speaking; the JSON standard itself defines some datatypes. It's a rule-based constrained structure which can be parsed into an object, and it can be validated. You'd probably only want to use it in the content model for an element, but with appropriate escaping it could also function as an attribute value, albeit a huge one.
I can't see this being an attribute value either.
I can actually see people putting GeoJSON into the @n attribute on <geo>, or (more likely) creating a new @geojson attribute for <geo> to contain it. There's no limit on the size of attribute values, after all.
It all depends on what we mean by a datatype. What does TEI mean by "datatype"?
We mean (informally speaking) an abstraction which defines the set of possible values for some attribute or attributes. Or so I believe.
So we're never going to constrain element content based on datatypes? Cheers, Martin