On 22 Nov 2007, at 13:30, Jerome Louvel wrote:
Hi Henry,Thanks for sharing these thoughts and pointers. I see two related use cases in the Restlet framework :- logging information about high-level requests (not solely in HTTP) comingin and going out of a Component, an Application. - logging information about low-level requests coming from HTTP server connectors or going through HTTP client connectors.
Logging is not really the most interesting bit here.Knowing where information has come from, keeping track of the headers is the first step to creating a good caching system.
If you look at tabulator you will see that this helps the tabulator work out when it needs to fetch representations for resources it has not got in its database. Since tabulator is a hyperdata browser, data that links to other data, that links to yet further data, there is always potentially more information to be found. The data found can be in the form of facts or ontologies (schemas is what this most closely resembles). So we have extensbible schemas and growable network of facts.
For those use cases, I would prefer to use a more compact XML syntax, but the ability to generate a RDF version would be interesting for exchange andmanipulation with other RDF infos.
Inventing new syntax's is usually a waste of time. I bet whatever someone come up with won't be that much more compact than some of the serialisations of rdf. I don't see how it could be more compact than N3 for example. It also won't be as flexible.
I've updated the related RFE: http://restlet.tigris.org/issues/show_bug.cgi?id=238
Ok. I'll let you know how I get along and what further thoughts my project brings me to.
Best regards, Jerome
all the best from Brazil,
Henry
smime.p7s
Description: S/MIME cryptographic signature
