HTTP is reuest/response based statelss protocol i.e No state is maintained across requests.
If we look definition of 'connection' in HTTP 1.1 RFC connection A transport layer virtual circuit established between two programs for the purpose of communication.
So connection based/connection less term is generally applied to transport levels protocols ( e.g TCP/IP is connection based while UDP is connectionless). HTTP is not a trasnport layer protocol so I think connection based/connection less does not make any sense. Applicable terms are statfeul/stateless.
HTTP 1.0 (Hyper Text Transfer Protocol) Port 80. Built on top of TCP/IP. Stateless, connection less and directional. Client opens a connection (using a URL � the direction) and sends request to the server, the server responds using the same connection, then the connection is closed. It is up to the client to optionally maintain the session (through cookies or URL re-writing).
HTTP 1.1 (Hypertext Transfer Protocol) Port 80. Same as HTTP 1.0 except there is an option to keep the connection alive from request to request. This would be known as connection oriented if it weren't for the fact that either side can potentially drop the connection at any time (Apache does after 15 seconds), as such it is still 'connection-less'. However, some people think it is connect oriented because you need a connection to be open to do a request at all. I disagree, but I get the impression this is what Sun thinks, I suppose that is what counts for the exam at least, it is very confusing.
William Butler Yeats: All life is a preparation for something that probably will never happen. Unless you make it happen.