openRDF.org Issue Tracker

[RIO-69] N3 and Turtle media types need updating

Tue, 7 Jul 2009 19:00:29 +0200 (CEST)

The MIME / media / content types for N3 and Turtle are currently defined as text/rdf+n3 and application/x-turtle respectively: http://repo.aduna-software.org/svn/org.openrdf/sesame/trunk/core/rio/api/src/main/java/org/openrdf/rio/RDFFormat.java .

They need to be updated to text/n3 and text/turtle. For Turtle see the latest specification at http://www.w3.org/TeamSubmission/turtle/#sec-mediaReg . The N3 specification is a bit buggy with regard to that, it switches between text/rdf+n3 and text/n3: http://www.w3.org/TeamSubmission/n3/ . However see also http://www.w3.org/blog/SW/2008/01/15/w3c_team_submission_on_n3_and_turtle , http://chatlogs.planetrdf.com/swig/2009-04-01.html#T14-00-16 and some more discussion on http://lists.w3.org/Archives/Public/semantic-web/2009May/0021.html .

Since RDFFormat accepts a collection of MIME types perhaps it would be even better to add the old ones as well so that the right parser can be selected for data served with the old types. For Turtle the type application/turtle could be added to such a collection as well, as it was the primary type mentioned on http://www.dajobe.org/2004/01/turtle/2007-11-20/#sec-mime .

[RIO-72] There is no way to specify "preserveBNodeIDs" property when doing RDF import

Tue, 1 Dec 2009 18:35:05 +0100 (CET)

* There is now way to pass pre-created to RepositoryConnection.add(Reader|InputStream|File) method with this option enabled.
* Both RDFParserBase and RDFInserter have preserveBNodeIDs property. In the first class it is false by default, in the second is true by default. There should be a single place for this functionality.

Nice to have feature:
- renumerateBNodeIDs property for RDFWriter, which should re-number nodes like "n1", "n2", "n3" ([see forum|http://sourceforge.net/mailarchive/forum.php?thread_name=4B139FB4.5080809%40mail.ru&forum_name=sesame-general])

[RIO-76] RdfParserRegistry prints error to logs when it doesn't find a built-in parser factory

Fri, 24 Jun 2011 13:22:19 +0200 (CEST)

I have only imported the rio packages that I require using Maven, however, there are errors printed for the missing packages, ie, Trig and Trix. These should be warnings or informational at most, as the package still operates properly without the missing parser factories.

See below for a stack trace.

ERROR (org.openrdf.rio.RDFParserRegistry:58) Failed to instantiate service
sun.misc.ServiceConfigurationError: org.openrdf.rio.RDFParserFactory: Provider org.openrdf.rio.trig.TriGParserFactory not found
at sun.misc.Service.fail(Service.java:129)
at sun.misc.Service.access$000(Service.java:111)
at sun.misc.Service$LazyIterator.next(Service.java:273)
at info.aduna.lang.service.ServiceRegistry.<init>(ServiceRegistry.java:42)
at info.aduna.lang.service.FileFormatServiceRegistry.<init>(FileFormatServiceRegistry.java:20)
at org.openrdf.rio.RDFParserRegistry.<init>(RDFParserRegistry.java:33)
at org.openrdf.rio.RDFParserRegistry.getInstance(RDFParserRegistry.java:26)
at org.openrdf.rio.Rio.createParser(Rio.java:167)
at org.openrdf.rio.Rio.createParser(Rio.java:188)
at org.openrdf.repository.base.RepositoryConnectionBase.addInputStreamOrReader(RepositoryConnectionBase.java:373)
at org.openrdf.repository.base.RepositoryConnectionBase.add(RepositoryConnectionBase.java:277)

I would prefer not to have to filter these messages, as I would like to see Sesame/Rio errors in my logs if they are legitimate.

[RIO-78] RDFFormat.register(RDFFormat) is not thread-safe

Mon, 27 Feb 2012 16:36:18 +0100 (CET)

This is obvious by inspection.

The backing data structure is an array list. That list is not wrapped with a Collections.synchronizedList().

private static List<RDFFormat> RDF_FORMATS = new ArrayList<RDFFormat>(8);

The register() method also does not use any synchronization.

public static void register(RDFFormat rdfFormat) {
RDF_FORMATS.add(rdfFormat);
}

As a consequence, concurrent modification exceptions are possible with this API.