Storage

• Node Storage
  • Requirements for NodeStorage Implementations
  • General Notes for Implementors
• Identifiers Used to Store Nodes
  • NodeBackendHandle (192 bit)
    • NodeBackendID (64 bit)
    • Free Tagging Bits (13 bit)
    • Inlinining tagging bit (1 bit)
    • RDFNodeType (2 bit)
    • NodeID (48 bit)
      • LiteralType (6 bit)
      • LiteralID (42 bit)
        • LiteralID with inlined language tag
  • Important Notes
• Backend Views

In rdf4cpp, we try our best to spare the user the details of how graphs and RDF nodes are actually stored in the backend. So if you just want to use rdf4cpp, you most probably do not need to go down this rabbit hole. You can just use the defaults.

Storage is split up into two parts, Node Storage and tuple storage (DatasetStorage). For both storage parts, reference implementations are provided and used as a sensible default.

NodeStorage

NodeStorages store information about RDF Nodes. By default, there is only one default NodeStorage which is automatically used (it lives at rdf4cpp::storage::default_node_storage). The constructors of BlankNode, IRI, Literal and Variable optionally allow to use another NodeStorage. The SyncReferenceNodeStorage implementation of the NodeStorage concept, which is used by default, is thread-safe. Whereas the UnsyncReferenceNodeStorage implementation is not. For more details, consult Node Storage.