WebFund 2016W Lecture 4

Video

The video for the lecture given on January 19, 2016 is now available.

Notes

In-class Notes

Lecture 4
---------
Web servers

What does a web server do?
Listens on port 80, 443, or other port
INPUT: HTTP requests
OUTPUT: HTTP responses & requested documents

tinywebserver.js
 - written in JavaScript (Node)
 - can only serve files
 - understands only a few file types

So tinywebserver is how web servers worked in 1995

What changed since then is now web servers run code to generate responses

Why use frameworks?
 - because we don't want to build up HTML from scratch
 - and, we want logic that is abstracted away
 - and, we want routing
   - have someone else decide which function to call in response to requests

But for this week, bare bones

Debugging

Student Notes

Announcements

• Assignment 1 due tomorrow (now pushed back to Thursday before class)
• Basics for tutorial 2 and assignment 2 are up!

Web Servers

• What does a web server do?
  • Listens on port 80 (usually HTTP), 443 (usually HTTPS) or some other port.
    • If the web server is listening on a non-standard port, you must specify the port in the URL (using something like “localhost:8080”)
  • Input: HTTP requests
  • Output: HTTP responses with requested documents
    • HTTP responses always have a response code
      • 200: OK, used for successful request
      • 404: Not Found
  • In other words it services requests
• Need root privileges to listen on ports <~1024
  • Do not do development as root if you can, it is risky!
• Popular Web Servers: Apache, NginX

Tiny Web Server

• Written in JavaScript(Node)
• Can only serve files! (most modern servers can do more)
• Understands limited file types
• Similar to web servers from 1995
• To try to understand what the code is doing, we can start by looking at the bottom of the file

http.createServer()

• Apart from the loading of some modules and the definitions of some objects and functions, the first real code to run in this script is the createServer() method
• We are calling a method of the http object (loaded from a module) in order to create an object which will act as the server (which we then store in the server variable)
• This method takes a single argument, request_handler
  • request_handler is a function which is intended to handler incoming requests
  • By providing this function as the argument, we are telling the server to call the function every time a request comes in

server.listen()

• Here we are calling the listen() method of the server object that was just created
• This tells the server to start listening for incoming requests
• The first two arguments are telling the server where to listen (on port 8080 of localhost)
  • This information came from the options object which was defined at the top of the file
• The third argument is a callback function
  • This will run once the server has started listening
  • It simply prints a message in the console to indicate that the server is ready

request_handler()

• This function gets called every time a request comes in (because we provided it as an argument to the http.createServer() method)
• There are two arguments
  • request is an object which represents the request that was received
  • response is an object which represents the response that will be sent
• At a high level, this function checks for the existence of a requested file and responds with either the file or an error code
• There are many nested callback functions here
  • Although this makes the code more difficult to read, why is this preferable to doing things synchronously?
    • This ensures that the request_handler() function will return very quickly
    • All the bulk of the work is done asynchronously
    • This allows the server to be available to handle another request again very quickly
    • This is important in order to keep the server responsive for numerous requests

MIME Types

• MIME types are used to specify the type of data being sent via HTTP
• In Tiny Web Server, there is an object MIME_TYPES defined near the top of the file which contains the types of files that the server recognizes
• The MIME_TYPES object is used in the get_mime() function to determine the type of a requested file
• Similarly, the MIME_TYPES object is used to set the Content-Type header of the HTTP response that is sent by the server
• We can break ‘image/jpg’ by changing it to ‘imaage/jpg’
  • The web browser does not know how to handle it because it does not recognize imaage
• Web browsers can deal with broken content type for some types of files
  • For example, putting a jpg as png in MIME_TYPES, will not break the content and Firefox is smart enough to render it correctly as jpg
  • To prevent the browser from attempting to detect what format the content is, the X-Content-Type-Options: nosniff header can be added.

Response Headers

• Some headers of the HTTP response are set in the respond() function
  • This is done using the response.writeHead() method
  • We could add more headers here if we want to
• Some other headers are set for us by the code in the http module
• We can see these headers from within the dev tools of the web browser after a response has been received
• There are not many headers used here because Tiny Web Server is not very complex

Frameworks and Modern Servers

How have things improved since 1995?

• We use code to generate responses to requests
  • We often use frameworks to do much of the work for us

Why use frameworks?

• Because we don’t want to build up HTML from scratch
• Abstracted away logic
• We also want routing (dispatcher): have someone else decide which function to call in response to requests

Debugging

• You can use the node debugger by typing node debug script.js in the terminal
• This will stop the program on the first line of code that is run
• You can continue execution of the code by typing c
• You can learn about other useful commands by typing help
• In particular, you can type repl to enter a read-evaluate-print loop from which you can inspect or modify objects or other data
• You can also write debugger; in the script to create a break point where the debugger will pause the execution of the program
• There is also a debugger available in the web browser
  • This is for client side code and will not work for any server-side code

HTML

• You can use the inspect in the web browser dev tools to see the the tree-structure HTML which makes up a web page
• <div> tags are used for sections in HTML
• <a> tags are used for links (review on CodeAcademy)