I'm not sure why the stack trace seems to be obscuring the actual RepositorySourceException message, but line 239 in the Graph.java class is throwing the exception because it is attempting and failing to obtain a connection to a repository source by name. If I had to guess, it'd be that the Graph is being created with a wrong name or with a name that doesn't match a repository source defined in the configuration file. Can you post the actual code in your JcrProviderImpl.init method?

Best regards,

Randall

Whoa, sorry 'bout that. My bad.

I assume you're using the same files as provided in the examples. If so, then all the interesting configuration content is given in the "configRepository.xml" file under the "/jcr:system" XML element, which after import would correspond to that same path in the "Configuration" repository.

Now, the RepositoryService connects to the repository named in the constructor and reads this repository to discover (among other things) the sources. And the RepositoryService starts looking for these nodes under a given path, which defaults to "/dna:system". So, unless another path is provided in the constructor, the RepositoryService will be looking under "/dna:system". But in the configuration repository, the path is "/jcr:system".

So, to correct it, simply use the other RepositoryService constructor [1] by providing "/jcr:system" as the path to the top of the configuration repository. That should get you past this problem. Please let us know if you run into other problems.

Clearly the example configRepository.xml file should be aligned with the default path assumed by the RepositoryService. I've logged this as a bug in JIRA [2], and we'll fix it before this week's 0.5 release. (BTW, the 0.5 release now has a vastly-improved configuration mechanism, allowing you to programmatically declare the configuration of the sources while all the service wiring is done for you. Stay tuned for that.)

Best regards,

Randall

[1] http://www.jboss.org/file-access/default/members/dna/freezone/docs/0.4/apidocs/org/jboss/dna/repository/RepositoryService.html#RepositoryService(org.jboss.dna.repository.RepositoryLibrary,%20java.lang.String,%20java.lang.String,%20org.jboss.dna.graph.property.Path,%20org.jboss.dna.graph.ExecutionContext)
[2] https://jira.jboss.org/jira/browse/DNA-436

The specific of acquiring the Credentials object is not specified by the JCR API, that means implementation may provide their own. Thus for now DNA JCR does not provide an Credentials implementation (there is an open discussion apropos).
When I take a look at this snippet code from info.magnolia.cms.util.WorkspaceAccessUtil

/**
 * Login to the specified repository/default workspace using given credentials
 * @param credentials
 * @param repositoryName
 * @return newly created JCR session
 * @throws RepositoryException if login fails or workspace does not exist
 * */
 public Session createRepositorySession(SimpleCredentials credentials,
 String repositoryName) throws RepositoryException {
 return this.createRepositorySession
 (credentials, repositoryName, ContentRepository.getDefaultWorkspace(repositoryName));
 }

The 0.5 release will also accept javax.jcr.SimpleCredentials as well, but it still uses JAAS under the covers. The trunk has some examples of how to configure this (q.v. dna-jcr/src/test/java/org/jboss/dna/jcr/InMemoryRepositoryStub.java IIRC)

We're trying to close out the release by the end of this week.