Two questions: Is there any technical reason jcr clients can't use the name[index] name pattern, similar to oak internal?
When importing content from jackrabbit which contains content with SNS, what't the recommended approach to edit such nodes? Now I'm getting exceptions like 20:38:35,833 ERROR services.RepositoryAdminService.addProperty() - line 331 [127.0.0.1] - unable to add property; javax.jcr.PathNotFoundException: /nen:content[2] at org.apache.jackrabbit.oak.jcr.session.SessionContext.getOakPathOrThrowNotFound(SessionContext.java:358) at org.apache.jackrabbit.oak.jcr.version.VersionManagerImpl.getOakPathOrThrowNotFound(VersionManagerImpl.java:248) at org.apache.jackrabbit.oak.jcr.version.VersionManagerImpl.access$000(VersionManagerImpl.java:53) Is it expected that all content imported from jackrabbit is to fixup all existing node names to avoid SNS and avoid using the name[index] name pattern, before such content can be modified in oak? On 5 August 2014 17:13, Michael Dürig <[email protected]> wrote: > > > On 5.8.14 2:56 , Torgeir Veimo wrote: >> >> So since that issue is unresolved, I assume I have to add the index >> manually. If I then use a naming scheme like name[index], I get an >> exception; >> >> javax.jcr.RepositoryException: Cannot create a new node using a name >> including an index > > > AFAIR you currently can't do that through the JCR API with Oak. This is only > supported on Oak's internal interfaces: "the underlying oak-core and > microkernel -level SNS nodes simply used the "name[index]" naming pattern > with no extra semantics associated with it." [1]. We added this so Oak can > cope with e.g. node type definitions, which make use of SNSs. > > Michael > > [1] https://issues.apache.org/jira/browse/OAK-203 -- -Tor
