Think about it. There is no hypertext, just Bson packages, with self described contents. Evan absent the semantic revolution, Bson alone has done away with Http. What's the point? May as well run high speed Bson streams straight though sockets.
Separately the semantic system will use the named graph system to manage the ip vs Url tables. That's has to be the overall plan, just dump http subsume the DNS indexing into the local named graph tables, let the web bots go about keeping them up to date. I think I am pursuing this approach with the semantic machine.
The other point is what's the point of a CGI when your server holds nothing but potentially executable Bson sub graphs? Just take the retreived object, convolve it with the user graph and execute the results, right there by calling the overload method in the same thread.
No comments:
Post a Comment