CSE 331 Software Design and Implementation

Slides:



Advertisements
Similar presentations
Chapter 16 Graphical User Interfaces John Keyser’s Modifications of Slides by Bjarne Stroustrup
Advertisements

OOP Design Patterns Chapters Design Patterns The main idea behind design patterns is to extract the high level interactions between objects and.
The Bridge Pattern.. Intent Decouple an abstraction from its implementation so that the two can vary independently Also known as: Handle/Body.
Liang, Introduction to Java Programming, Eighth Edition, (c) 2011 Pearson Education, Inc. All rights reserved Immutable Objects and Classes.
System Architecture Lecture 3 CSE 111 Spring /22/20151Copyright William E. Howden.
Aalborg Media Lab 23-Jun-15 Inheritance Lecture 10 Chapter 8.
Threads II. Review A thread is a single flow of control through a program Java is multithreaded—several threads may be executing “simultaneously” If you.
Event Handling in Java: Alternatives and Patterns Raja Sooriamurthi Information Systems Department Kelley School of Business Indiana.
REFACTORING Lecture 4. Definition Refactoring is a process of changing the internal structure of the program, not affecting its external behavior and.
Chapter 3 Introduction to Collections – Stacks Modified
Observer Design Pattern Source: Design Patterns – Elements of Reusable Object- Oriented Software; Gamma, et. al.
Software Construction Lecture 10 Frameworks
Implementing Design Patterns Using Java St. Louis Java Special Interest Group Eric M. Burke Object Computing, Inc. Presented on July 9, 1998 (updated July.
CSC 313 – Advanced Programming Topics. Observer Pattern Intent  Efficiently perform 1-to-many communication  Easy to respond dynamically when event(s)
Design Patterns Lecture III. What Is A Pattern? Current use comes from the work of the architect Christopher Alexander Alexander studied ways to improve.
CSE 331 Software Design & Implementation Dan Grossman Winter 2014 Events, Listeners, and Callbacks (Based on slides by Mike Ernst, David Notkin, Hal Perkins)
CSE 331 SOFTWARE DESIGN & IMPLEMENTATION MODULAR DESIGN PRINCIPLES Autumn 2011 Google image search on “designing software modules”
Chapter 26 GoF Design Patterns. The Adapter Design Pattern.
CSC 313 – Advanced Programming Topics. Design Pattern Intent  Each design pattern is a tool  Like all tools, have reason for being.
CSE 331 SOFTWARE DESIGN & IMPLEMENTATION MIDTERM REVIEW Autumn 2011.
3-1 State Design Pattern C Sc 335 Rick Mercer. State Design Pattern State is one of the Behavioral patterns It is similar to Strategy Allows an object.
Frameworks CompSci 230 S Software Construction.
CSC 313 – Advanced Programming Topics. Observer Pattern in Java  Java ♥ Observer Pattern & uses everywhere  Find pattern in JButton & ActionListener.
Liang, Introduction to Java Programming, Sixth Edition, (c) 2005 Pearson Education, Inc. All rights reserved Chapter 27 JavaBeans and.
OBSERVER DESIGN PATTERN. Behavioral Patterns  Behavioral patterns are those patterns that are most specifically concerned with communication between.
Design Patterns Software Engineering CS 561. Last Time Introduced design patterns Abstraction-Occurrence General Hierarchy Player-Role.
Concordia University Department of Computer Science and Software Engineering Click to edit Master title style ADVANCED PROGRAMMING PRACTICES Model View.
(1) Introduction to Java GUIs Philip Johnson Collaborative Software Development Laboratory Information and Computer Sciences University of Hawaii Honolulu.
1 CSE 331 Model/View Separation and Observer Pattern slides created by Marty Stepp based on materials by M. Ernst, S. Reges, D. Notkin, R. Mercer, Wikipedia.
The Mediator Pattern (Behavioral) ©SoftMoore ConsultingSlide 1.
CSE 331 Software Design & Implementation Hal Perkins Winter 2013 Events, Listeners, and Callbacks (slides by Mike Ernst and David Notkin) 1.
1 COS 260 DAY 12 Tony Gauvin. 2 Agenda Questions? 5 th Mini quiz –Chapter 5 40 min Assignment 3 Due Assignment 4 will be posted later (next week) –If.
L10: Model-View-Controller General application structure. User Interface: Role, Requirements, Problems Design patterns: Model – View – Controller, Observer/Observable.
The Observer Pattern (Behavioral) ©SoftMoore ConsultingSlide 1.
Observer / Observable COMP 401 Fall 2014 Lecture 14 10/7/2014.
Slide design: Dr. Mark L. Hornick
Observer Pattern Keeping An Eye on Things Need to introduce observer pattern formally first, include book definition & design principle Keeping An Eye.
Module dependences and decoupling CSE 331 University of Washington.
Observer Pattern Context:
Design Patterns Source: “Design Patterns”, Erich Gamma, Richard Helm, Ralph Johnson, John Vlissides And Created.
Abstract Factory Pattern
SE-2811 Software Component Design
Creational Pattern: Prototype
Observer Design Pattern
Observer Design Pattern
Architectural Patterns for Interactive Software
Abstract Factory Pattern
CSE 331 Software Design and Implementation
Event Driven Programming
CSE 331 Software Design and Implementation
Design Patterns, cont..
Advanced Programming Behnam Hatami Fall 2017.
CSE 331 Software Design and Implementation
Introduction to Behavioral Patterns (1)
CSE 331 Software Design and Implementation
CSE 143 Lecture 2 Collections and ArrayIntList
Behavioral Design Pattern
CSE 331 Software Design & Implementation
Design pattern Lecture 9.
Advanced ProgramMING Practices
8. Observer Pattern SE2811 Software Component Design
9. Threads SE2811 Software Component Design
Advanced ProgramMING Practices
Model, View, Controller design pattern
9. Threads SE2811 Software Component Design
5. Strategy, Singleton Patterns
CSE 331 Software Design & Implementation
Software Design Lecture : 40.
Software Design Lecture : 28.
9. Threads SE2811 Software Component Design
Presentation transcript:

CSE 331 Software Design and Implementation Lecture 16 Events, Listeners, Callbacks Zach Tatlock / Spring 2018

The limits of scaling What prevents us from building huge, intricate structures that work perfectly and indefinitely? Not just friction Not just gravity Not just wear-and-tear … it’s the difficulty of managing complexity! So we split designs into sensible parts and reduce interaction among the parts More cohesion within parts Less coupling across parts

Design exercise #1 Write a typing-break reminder program Offer the hard-working user occasional reminders of the perils of Repetitive Strain Injury, and encourage the user to take a break from typing.

Design exercise #1 Write a typing-break reminder program Offer the hard-working user occasional reminders of the perils of Repetitive Strain Injury, and encourage the user to take a break from typing. Naive design: Make a method to display messages and offer exercises Make a loop to call that method from time to time (Let's ignore multithreaded solutions for this discussion)

TimeToStretch suggests exercises public class TimeToStretch { public void run() { System.out.println("Stop typing!"); suggestExercise(); } public void suggestExercise() { …

Timer calls run() periodically public class Timer { private TimeToStretch tts = new TimeToStretch(); public void start() { while (true) { ... if (enoughTimeHasPassed) { tts.run(); }

Main class puts it together class Main { public static void main(String[] args) { Timer t = new Timer(); t.start(); } This program, as designed, will work... But we can do better

Module dependency diagram (MDD) An arrow in a module dependency diagram (MDD) indicates “depends on” or “knows about” Simplistically: “any name mentioned in the source code” What’s wrong with this diagram? Does Timer really need to depend on TimeToStretch? Is Timer re-usable in a new context? Main Main depends on Timer Timer TimeToStretch Timer depends on TimeToStretch

Decoupling Timer needs to call the run method Timer does not need to know what the run method does Weaken the dependency of Timer on TimeToStretch Introduce a weaker specification, in the form of an interface or abstract class public abstract class TimerTask { public abstract void run(); } Timer only needs to know that something (e.g., TimeToStretch) meets the TimerTask specification

TimeToStretch (version 2) public class TimeToStretch extends TimerTask { public void run() { System.out.println("Stop typing!"); suggestExercise(); } public void suggestExercise() { ...

Timer (version 2) public class Timer { private TimerTask task; public Timer(TimerTask task) { this.task = task; } public void start() { while (true) { ... task.run(); Main creates a TimeToStretch object and passes it to Timer: Timer t = new Timer(new TimeToStretch()); t.start();

Module dependency diagram (version 2) Timer depends on TimerTask, not TimeToStretch Unaffected by implementation details of TimeToStretch Now Timer is much easier to reuse Main depends on the constructor for TimeToStretch Main still depends on Timer (is this necessary?) Main TimerTask Timer Dependence TimeToStretch Subclassing

The callback design pattern An alternative: use a callback to invert the dependency TimeToStretch creates a Timer, and passes in a reference to itself so the Timer can call it back This is a callback – a method call from a module to a client that it notifies about some condition The callback inverts a dependency Inverted dependency: TimeToStretch depends on Timer (not vice versa) Less obvious coding style, but more “natural” dependency Side benefit: Main does not depend on Timer

Callbacks Callback: “Code” provided by client to be used by library In Java, pass an object with the “code” in a method Synchronous callbacks: Examples: HashMap calls its client’s hashCode, equals Useful when library needs the callback result immediately Asynchronous callbacks: Examples: GUI listeners Register to indicate interest and where to call back Useful when the callback should be performed later, when some interesting event occurs

TimeToStretch (version 3) public class TimeToStretch extends TimerTask { private Timer timer; public TimeToStretch() { timer = new Timer(this); } public void start() { timer.start(); public void run() { System.out.println("Stop typing!"); suggestExercise(); ... Register interest with the timer Callback entry point

Main (version 3) TimeToStretch tts = new TimeToStretch(); tts.start(); Uses a callback in TimeToStretch to invert a dependency This MDD shows the inversion of the dependency between Timer and TimeToStretch (compare to version 1) Main does not depend on Timer TimeToStretch depends on Timer Main TimerTask Timer TimeToStretch

Decoupling and design A good design has dependences (coupling) only where it makes sense While you design (before you code), examine dependences Don’t introduce unnecessary coupling Coupling is an easy temptation if you code first Suppose a method needs information from another object: If you hack in a way to get it: The hack might be easy to write It will damage the code’s modularity and reusability More complex code is harder to understand

Design exercise #2 A program to display information about stocks Stock tickers Spreadsheets Graphs Naive design: Make a class to represent stock information That class updates all views of that information (tickers, graphs, etc.) when it changes

Module dependency diagram Main class gathers information and stores in Stocks Stocks class updates viewers when necessary Problem: To add/change a viewer, must change Stocks Better: insulate Stocks from the vagaries of the viewers Main Stocks StockTicker Spreadsheet StockGraph

Weaken the coupling What should Stocks class know about viewers? Only needs an update method to call with changed data Old way: void updateViewers() { ticker.update(newPrice); spreadsheet.update(newPrice); graph.update(newPrice); // Edit this method to // add a new viewer.  }

Weaken the coupling CSE331 Fall 2015 What should Stocks class know about viewers? Only needs an update method to call with changed data New way: The “observer pattern” interface PriceObserver { void update(PriceInfo pi); } class Stocks { private List<PriceObserver> observers; void addObserver(PriceObserver pi) { observers.add(pi); void notifyObserver(PriceInfo i) { for (PriceObserver obs : observers) obs.update(i); … Register a callback Do the callbacks CSE331 Fall 2015

The observer pattern Stocks not responsible for viewer creation Main passes viewers to Stocks as observers Stocks keeps list of PriceObservers, notifies them of changes Issue: update method must pass enough information to (unknown) viewers Create viewers and get observers Main Create Stocks and add observers PriceObserver Stocks update StockTicker Create (or be) observers Spreadsheet StockGraph

A different design: pull versus push The Observer pattern implements push functionality A pull model: give viewers access to Stocks, let them extract the data they need “Push” versus “pull” efficiency can depend on frequency of operations (Also possible to use both patterns simultaneously.) new(Stocks) Main Stocks.new Stocks StockTicker Spreadsheet StockGraph

Another example of Observer pattern // Represents a sign-up sheet of students public class SignupSheet extends Observable { private List<String> students = new ArrayList<String>(); public void addStudent(String student) { students.add(student); setChanged(); notifyObservers(); } public int size() { return students.size(); … Part of the JDK SignupSheet inherits many methods including: void addObserver(Observer o) protected void setChanged() void notifyObservers()

cast because Observable is not generic  An Observer Part of the JDK public class SignupObserver implements Observer { // called whenever observed object changes // and observers are notified public void update(Observable o, Object arg) { System.out.println("Signup count: " + ((SignupSheet)o).size());‏ } Not relevant to us cast because Observable is not generic 

Registering an observer SignupSheet s = new SignupSheet(); s.addStudent("billg"); // nothing visible happens s.addObserver(new SignupObserver()); s.addStudent("torvalds"); // now text appears: "Signup count: 2" Java's “Listeners” (particularly in GUI classes) are examples of the Observer pattern (Feel free to use the Java observer classes in your designs – if they are a good fit – but you don’t have to use them)

User interfaces: appearance vs. content It is easy to tangle up appearance and content Particularly when supporting direct manipulation (e.g., dragging line endpoints in a drawing program) Another example: program state stored in widgets in dialog boxes Neither can be understood easily or changed easily This destroys modularity and reusability Over time, it leads to bizarre hacks and huge complexity Code must be discarded Callbacks, listeners, and other patterns can help