Persistence Layer

[npilon]

The Persistence Layer needs to maintain a set of graphs and manage links between graphs, addition of new graphs, and recording of updated graphs.

[npilon]

My thoughts here are as follows:

1. The persistence layer needs to manage a set of graphs. Each graph needs to be flagged as being modifiable (and, if modifiable, have the information needed to modify it) or read-only (IE, generated from a construct query).
2. When attempting to classify a Resource that has no type information in the current graph, Resource needs to ask the persistence layer if it knows of a graph that has type information. Once such a graph is found, it should be the graph for the new Resource.
3. How the persistence layer finds a graph needs to be customizable. Retrieving one by HTTP or SPARQL 1.1 Uniform HTTP Protocol is one possibility, but a SPARQL CONSTRUCT or DESCRIBE query could also make sense.
4. A modifiable graph should understand how to update itself. It should be possible to "commit" changes to one graph or all graphs.
5. Resources write into a single, specific graph. By default, this should be the graph that contains that Resource's type information. It should be possible to specify another graph to write to.
6. I'm of the opinion that reads should either be from the default graph or require a specified graph. But that's not a particularly firm opinion.