ITEC 325 2017fall ibarland

home—lects—hws
D2L—breeze (snow day)

client-and-server-2
intro: web requests and responses

youtube: inside http headers (19m55s)

Recall: In essence, a web-server and web-browser are each functions with the type-signatures:
server: string_url → string_html
browser: string_html → void.

Looking more closely

For this class, you'll need a browser with good debug-tools built in. For Firefox, there is an extension “Firebug”; I'll demo a couple of Chrome's built-in tools. We will not worry about supporting old browers (e.g. old versions of IE).

≡ » More Tools » Developer Tools
We’ll look briefly at the Elements tab, but today we're mostly concerned with the Network tab, and clicking on particular files. (You might need to click on View » Hide Overview.) In particular, viewing the Request Headers and Response Headers.

Together in class: Spend 15min looking at examples.

Btw, here are the official HTTP response codes, or the cat-picture version.

the real upshot: So, the essence of the situation is actually that the server and browser are functions with the type-signatures:
server: http-request-packet → http-response-packet (where the first packet contains the URL, and the second packet contains the html response);
browser: http-response-packet → void.

Further things to consider (if not today, then perhaps later:)

• starting up at Starbucks, when you get 'welcome to our wireless; please agree to our terms'.
• telling my phone to use RUWireless, even not in browser a page comes up w/ "please authenticate"
• Q: are URLs case-sensitive? How might you experiment, to find out?
• What is the file that gets looked up, by the www.radford.edu web server? By php.radford.edu ?
• sites that pop up 20 other windows (usually adverts)? Whose fault is that?
• Referring page is often provided, when making a request. How is this a security leak?
  An actual example: Suppose a hiring committee has a wiki with their comments, and use “privacy through obfuscation”: they don't set up a password, but they do use an un-guessable URL.
• When you start typing a google-request, it auto-completes. What's happening? (What info is being sent, and when?)

        - ACTUAL: server-side computation (via Ajax):
               on each keystroke, a transaction is happening, and google.com computes the completion
               (for *every* google visitor in the world, yikes, is that a lot of bandwidth?)
  
        - What are some other *hypothetical* was this could be handled?
          - client-side: the server includes the top 10,000 most popular searches,
              and the client javascript uses that to determine the auto-completion.
              (This wouldn't explain the .)
  
            That would be an expensive load-time on every page!  How might we use this approach but
              try to avoid having 'google.com' be a huge download every time?
  
            How about: only include that dictionary on the *first* visit to google.com,
              and have the client javascript write it to a file;
              then on subsequent visits this
  
            This requires something additional though: How does google know if this is the *first*
              time I'm visiting, or if I've visited previously and therefore I already have 
              the set of common-searches cached?
              We'll discuss cookies later, which solve this.
          
            

• How do ad blockers work? ad blockers attack, and counter-attacks

home—lects—hws
D2L—breeze (snow day)

©2017, Ian Barland, Radford University Last modified 2017.Aug.28 (Mon)

Please mail any suggestions (incl. typos, broken links) to ibarlandradford.edu