Content-type headers such as text/anot+turtle affect content negotiation

[DylanVanAssche]

I have been fetching ontologies from OGC such as GeoSPARQL (http://www.opengis.net/ont/geosparql) or Simple Features (http://www.opengis.net/ont/sf). OGC uses prez for providing them. I see that prez injects some additional triples though, together with the ontology.

I have noticed that prez doesn't use the text/turtle as Content-Type, even though an Accept header is provided as Accept: text/turtle. I wonder what the reasoning behind text/anot+turtle is? AFAIK this is not a registered media type.

Related to rubensworks/rdf-dereference.js#64

[nicholascar]

@DylanVanAssche, the W3C has now initiated an update to ConnegP - https://www.w3.org/TR/dx-prof-conneg/ - the standard Prez is meant to follow for its API implementation.

Since ConnegP's first W3C release, I expect Prez has drifted a bit from the specification so likely we - Prez developers - will have to fix up a few small things.

The issue you are specifically calling out - the use of text/anot+turtle - will certainly need improving. Prez adds in ll the labels it can find for all the Subjects, Predicates & Objects in the RDF graph that is the actual data package that the particular Prez endpoint returns. This helps non-Semantic Web formats like JSON present nicely labelled key/value pairs of information and allowd PrezUI to make nice looking web pages for things. But it is tricky as you say, since text/anot+turtle is an un-registered Media Type.

text/anot+turtle responses are still valid text/turtle, so you should be able to use them in any system expecting RDF, you will just get more labels.

You will see some ConnegP issues forming here in Prez' tracker shortly and I hope updates to the OGC Reference Data Platform will occur as soon as those updates have been merged in.

One set of updates we already have in place elsewhere is a better listing of the available formats and profiles for a resource. With that update in place, you should more easily be able to choose to get un-altered GeoSPARQL/SL or the annotated forms.

[DylanVanAssche]

@nicholascar

Thanks for explaining in depth the reasoning behind the text/anot+turtle, it gives a lot of insight. I will check with https://github.com/rubensworks/rdf-dereference.js on how to proceed further in the tooling to handle this.