9/203.3.4 HTTP
The term Hypertext Transfer Protocol (HTTP) is perhaps misnamed because HTTP is instead a transfer protocol for any information with the capability for making hypertext jumps [BLCL+94]. HTTP is based upon TCP/IP like other Internet protocols (FTP, NNTP), but in contrast to those, it is stateless. A TCP/IP connection is opened to transfer or manipulate data, and closed as soon as one single operation is finished. This can result in several HTTP connections to retrieve a single document, for instance, if that document contains graphics because each retrieval of an image is one operation.A HTTP request from a client consists of a command (``GET'', ``PUT'', ``POST''), the URL of the object which is concerned, and, if needed, additional data. The server replies with the requested object, an acknowledgment of the manipulation, or an error message (the most common error is 404 Not Found). When an object is transfered, a HTTP-header, which contains metainformation about the object and the browser, is attached. This allows the transfer of any kind of data because the browser gets the needed information about the data in the HTTP-header. Since unrecognised information in the header is ignored by the server, the protocol can easily be extended.