I’ve decided to attempt a REST-ful tuple space implementation for RDF triples, thereby increasing my buzzword concatenation quotient to previously undreamt-of levels.
Last night I sorted out some of the basics: getting MySQL and phpMyAdmin set up, building a couple of tables, getting a working connection in some PHP code. OK, OK, it’s accessible…
Because tuple space read operations are blocking, and HTTP is a synchronous protocol, the client may submit a URL to which a notification can be PUT when a tuple becomes available. The rough semantics is as follows:
Client POSTs a request to create a new tuple space to the server. This creates a resource representing the tuple space. A GET on this resource will return a list of resources representing the tuples in the tuples space at that time. A DELETE will delete the tuple space.
Client POSTs a tuple into a tuple space. This creates a resource representing that tuple. The resource can be accessed using GET and deleted with DELETE.
Client POSTs a request for a tuple (and optionally a notification URL) to a resource representing a tuple space’s
Andrew Rowstrom’s
I think that should cover it. By restricting the tuples to be RDF triples, we simplify the pattern matching and storage requirements quite considerably. Our tuple space has just become a triple store! Boundless good things will undoubtedly ensue.