Clients and Servers

1
Clients and Servers
2
URL review

• A URL has the syntax
• protocol//hostnameport/pathanchor
• import java.net.
• This is the package that defines sockets, URLs,
  etc.
• URL url new URL(String)
• Constructs a URL object from a text string
• MalformedURLException
• This exception is thrown if the given String
  cannot be parsed by newURL(String)
• We have used URLs to display a page in an applet
• appletContext.showUrl(URL)

3
HTTP review

• HTTP is a protocol--a formal description of a
  language that computers use to communicate
• An HTTP message consists of three parts
• The request or the response line
• A request line typically contains either GET or
  PUT
• A response line contains the status code, such as
  404 Not Found
• A header section
• Contains name-value pairs, such as Content-type
  text/html
• Ends with a blank line
• The body of the message
• The body is optional

4
Using a URL

• URLConnection c url.openConnection()
• The URLConnection is the basic way to access the
  resource information
• c.getHeaderField(name)
• Returns the value of the named header field (as a
  String)
• Frequently used fields have shorthand methods,
  for example,c.getLastModified()
  c.getHeaderField("last-modified")
• getHeaderField(int)
• Returns the value of the int-th header field (as
  a String)
• The 0-th header field is the status line
• c.getInputStream()
• Returns an InputStream containing the content
  of the resource
• url.openStream() is shorthand for
  url.openConnection().getInputStream()

5
Socket review

• A socket is a low-level software device for
  connecting two programs (possibly on different
  computers) together
• new Socket(String host, int port)
• Creates a client socket and makes the connection
• Methods include getInputStream(),
  getOutputStream(), and close()
• new ServerSocket(int port)
• Creates a server socket that listens on the
  specified port
• accept() returns a Socket that can be used for
  I/O
• accept() is a blocking method, so multithreading
  is highly desirable

6
How to write a server

• ServerSocket server new ServerSocket(port)
• The port should be a number above 1024
• Socket client server.accept()
• accept() blocks while it waits for a connection
• InputStream inStream client.getInputStream()
• InputStreamReader reader new InputStreamReader(i
  nStream)
• BufferedReader input new BufferedReader(reader)
  
• char ch input.read(), String s
  input.readLine()
• OutputStream outStream client.getOutputStream()
  
• PrintWriter output new PrintWriter(outStream,
  true)
• true is so that you auto-flush, that is, dont
  fill the buffer
• output .print(X), output .println(X), output
  .println()
• input.close(), output.close(), server.close(),
  client.close()

7
How to write a client

• Socket server new Socket(ip_address, port)
• The ip_address can be the String "localhost"
• This method makes the actual connection
• InputStream inStream server.getInputStream()
• As on the previous slide
• OutputStream outStream server.getOutputStream()
  
• As on the previous slide
• input.close(), output.close(), server.close()
• As on the previous slide

8
How to write an HTTP server

• An HTTP server is just a server that follows the
  HTTP protocol (request/status line, header, blank
  line, body)
• Since HTTP is a text-based protocol, compliance
  is easy
• There are two versions of HTTP 1.0 and 1.1
• HTTP 1.0 is simpler and should be used if the
  special features of 1.1 are not required
• The most important change in HTTP 1.1 is that it
  can accommodate proxy servers
• The client and server must agree which version of
  HTTP is being used
• Most HTTP servers can use both

9
Proxy servers

• Proxies are important because they allow more
  than one server to use the same IP address
• There arent enough IP addresses to go around
• If you have a lot of clients, you need a lot of
  servers--but the user should not have to try
  multiple IP addresses

10
Multithreading

• server.accept() is a blocking call--Java stops
  and waits for a response before it continues
• This is only acceptable if the server never has
  more than one client
• A server needs to have a separate thread for each
  client
• There are two ways to create a Thread
• Write a class that extends Thread
• Override the public void run() method
• Create an instance of your class and call its
  (inherited) start() method
• Write a class that implements Runnable
• Implement the public void run() method
• Create an instance of your class
• Create a Thread object with this instance as a
  parameter to the constructor
• Call the Thread objects start() method

11
Synchronization

• While an object is being modified by one thread,
  no other thread should try to access it
• This leads to unpredictable (and difficult to
  debug) results
• You can synchronize an object
• synchronized (obj) code that uses/modifies obj
  
• synchronized is a statement type, like if or
  while
• No other code that is synchronized on this object
  can use or modify the object at the same time
• You can synchronize a method
• synchronized void addOne(arg1, arg2, ...) code
  
• synchronized is a method modifier, like public or
  abstract
• Only one synchronized method in a class can be
  used at a time (but this doesnt restrict other,
  non-synchronized methods)
• Synchronization can really hurt efficiency (and
  response time)
• It can be very difficult to make a program both
  safe and efficient

12
A synchronization analogy

• Imagine that you have a building with two
  entrances
• One entrance is always kept locked, and has a
  single key
• The other entrance is never locked
• Synchronization is like this
• Code that always uses the locked (synchronized)
  entrance has to wait for other code to exit and
  hand over the key
• Any code that uses the unlocked entrance can go
  into the building at any time, regardless of what
  other code may be there
• Thus, synchronizing code only protects you from
  other synchronized code!

13
The End