Tutorial doesn't work out-of-the-box about cliopatria HOT 4 OPEN

The error did go away when I added

:- use_module(library(semweb/rdfa)).

to lib/semweb/rdf_file_type.pl

from cliopatria.

@firefrorefiddle Thanks for reporting.

The Pirates demo loads data from a Turtle manifest file, which includes the following URL as a data source: http://semanticweb.cs.vu.nl/2009/11/sem/

The URL exposes an RDF/XML file that (erroneously) claims to be an HTML file in its Content-Type header (Media Type text/html). Library semweb/rdf_db does not implement enough clever heuristics in order to figure that the claim made in the HTTP reply header (=HTML) is inconsistent with the actual content of the HTTP reply body (=RDF/XML).

The error that you encountered is thrown later, when the HTML Media Type is matched onto the RDFa serialization format, whose library is not loaded by ClioPatria:

rdf_content_type('text/html', 0.66, rdfa). %AC

IMO two fixes are needed:

1. rdf_content_type/3 needs to be made into a multi-file predicate that is distributed over the RDF/XML, RDFa, Turtle/TriG, and N-Quads/N-Triples modules.
2. semweb/rdfa needs to be loaded by ClioPatria, because a user may wish to load data in that serialization format.

@JanWielemaker Do you agree?, especially WRT (1)?

from cliopatria.

It is not the whole story. The server serves the documentation if the accept type contains html, but the documentation is not in rdfa. So, after loading the rdfa plugin it will load the html and zero triples. Seems the critical fix is to fix the content negotiation at the server end (an nginx server).

from cliopatria.

Fixed the server, so at least this works again:

?- rdf_load('http://semanticweb.cs.vu.nl/2009/11/sem/').

The other stuff is probably a good idea, though I'm still a little in doubt.

from cliopatria.