HTTP

a request-response protocol

client is a user agent (browser), server is an application process running on a different computer

URI: uniform resource identifier

• URN: uniform resource name
  • unique name for resource (like a book’s ISBN)
  • relies on some authority to provide information
• URL: uniform resource locator
  • location on the web
  • expect HTTP GET request to provide information about this resource
• a modern form is IRI (international resource identifier) which can also include unicode characters, not just ASCII

URLs: http://www.example.org:5678/a/b.txt?tc=win&r=10#para5

• contain
  • authority
    • host (FQDN, fully qualified domain name)
    • port
  • request-URI
    • path
    • query
  • fragment
• browser connects to authority over TCP
• request-URI included in start line (/ is default)
• fragment is not sent to server, used to scroll user’s view

Content negotiation

• multiple representations via the same URI
• client-server negotiation determines what shows up

Request methods

• GET: get documents, no body in request
• POST: e.g. when you click submit, form information is included in the boyd
• HEAD: requests only the header fields to be returned

Request example:

GET /test.html HTTP/1.1
HOST: [www.example.org](http://www.example.org)
…

Response example:

HTTP/1.1 200 OK
Date: <timestamp>

Header fields are included in the requests, like:

• user agent
• referrer
• content type
• acceptable MIME types (data types like text/html, image/png, vide/mp4)
• character encoding (most popular being UTF-8 encoding of unicode)

HTTP response codes:

• three-digit number
• first digit is the class
  • 1: information
  • 2: success
  • 3: redirect
  • 4: client fucked up
  • 5: server fucked up
• other two digits are extra information

HTTP servers:

• initially, there weren’t that many, mostly FTP in the 90s

• then, NCSA made httpd for free, and promptly fucked off to work on Netscape (stopping all support)

• so developers started making patches, and turned it into “a patchy server”…the Apache server

• a main loop for a server looks like so:

  while forever
  listen on TCP port 80
  read request
  send response
  

• the server doesn’t know its own hostname, because host names may be on the same machine (virtual hosts) and different responses may be needed for different hostnames

• the server uses a config file to determine where “/” is (the document root)