HTTP is becoming the dominant method of communcation on the planet Earth. [Talking is still #1, TV is probably #2.] Protocols underlying HTTP (e.g., TCP/IP) and above HTTP (e.g., HTML) are are entrenched but evolving (e.g., non-legacy ATM and XML) ... but it's hard to imagine a world without HTTP. HTTP is an enabling technology for distributed collaborative information systems, which are now close to the core of human intellectual evolution.
HTTP is not terribly complicated in theory or practice. It imposes just enough organization to provide reliable, efficient, architecture-independent communication of objects encoded as byte streams in the real world. Even so, these are enormously important advantages over, e.g., ASCII streams.
Webservers and browsers are the ubiquitous, HTTP-talking programs that make the world-wide-web work. The benefits of HTTP communication can be used by other kinds of programs to achieve robustness, scalability, interoperability, and supportability. Most of the programs we write should be all those things, so shouldn't most of them be talking HTTP?
A Daylight-style oopish toolkit which provides generic HTTP support has been designed and implemented as part of the fedora development project. A variety of applications have been implemented using this toolkit. This presentation will discuss lessons learned from a year's experience with the HTTP program model as well as expectations for the future.
Oops, I forgot to mention that HTTP is a misnomer for "hypertext transfer protocol".