Presentation is loading. Please wait.

Presentation is loading. Please wait.

OO Design Principles Copyright © Vyacheslav Mukhortov, Nikita Nyanchuk-Tatarskiy, 2001-2004 Copyright © INTEKS LLC, 2003-2004.

Similar presentations


Presentation on theme: "OO Design Principles Copyright © Vyacheslav Mukhortov, Nikita Nyanchuk-Tatarskiy, 2001-2004 Copyright © INTEKS LLC, 2003-2004."— Presentation transcript:

1 OO Design Principles Copyright © Vyacheslav Mukhortov, Nikita Nyanchuk-Tatarskiy, 2001-2004 Copyright © INTEKS LLC, 2003-2004

2 OO Designer Activities  Class design  Package design  Describing the design patterns

3 Class design principles  ORR One Responsibility Rule  LSP Liskov Substitution Principle  LoDLaw of Demeter  OCPOpen-Closed Principle  ISPInterface Segregation Principle

4 ORR - One Responsibility Rule A class has a single responsibility: it does it all, it does it well, it does it only - R. Martin

5 LSP – Liskov Substitution Principle Functions that use pointers or references to base classes must be able to use objects of the derived classes without knowing it. - R.Martin, 1996 Original formula: If for each object o1 of type S there is an object o2 of type T such that for all programs P defined in terms of T the behavior of P is unchanged when o1 is substituted by o2 then S is a subtype of T. - Barbara Liskov, 1988

6 LSP violation: Rectangle & Square Problem: Square s = new Square(5); s.setHeight(6); // s is not consistent class Rectangle { private int h; private int w; public Rectangle( int w, int h ) { this.h = h; this.w = w; } public void setHeight( int h ) { this.h = h; } public int getHeight() { return h; } } class Square extends Rectangle { public Square( int s ) { super( s, s ); } }

7 class Square extends Rectangle { public Square( int s ) { super( s, s ); } public void setSize( int s ) { super.setHeight(s); super.setWidth(s); } public void setHeight( int h ) { setSize(h); } public void setWidth( int w ) { setSize(w); } } Problem: void f( Rectangle r ) throws Exception { r.setHeight(4); r.setWidth(5); if( r.getHight() * r.getWidth() != 20 ) throw new Exception( “Bug!” ); }

8 LSP: The real problem ? Square object is not a Rectangle object! Why? Because of behavior of a Square is not consistent with the behavior of a Rectangle! and … it is behavior that software is really all about! IsA is a behavioral relationship.

9 LoD – Law of Demeter Original formula: Only talk to your immediate friends. - Ian Holland, 1987 immediate friends of method f : methods of class of f and other argument classes of f methods of immediate part classes of class of f methods of classes of objects that are created in f. A method should have limited knowledge of an object model. - D. Rumbaugh

10 Grady Booch about The LoD “The basic effect of applying this Law is the creation of loosely coupled classes, whose implementation secrets are encapsulated. Such classes are fairly unencumbered, meaning that to understand the meaning of one class, you need not understand the details of many other classes.” You can play with yourself. You can play with your own toys (but you can't take them apart), You can play with toys that were given to you. And you can play with toys you've made yourself. More on LoD Peter Van Rooijen simple formula

11 LoD violation Problem: public void getTimeOfBirth() { long time = p.getDateOfBirth().getTime(); } Do not reveal a class secret!

12 LoD-compliant design Solution: void m() { this.b.call_foo(); } // hide the class secret!

13 OCP – Open-Closed Principle Software entities (classes, modules, functions, etc) should be open for extension, but closed for modification. - B. Meyer, 1988

14 OCP violation: Shapes void drawShapes( Shape[] shapes ) { for( int i = 0; i < shapes.length; ++i ) { if( shape[i].getType == Shape.SQUARE ) { drawSquare( (Square)shape[i] ); } else drawCircle( (Circle)shape[i] ); } The problem: You can't add a new shape without changing drawShapes() code.

15 OCP compliant solution void drawShapes( Shape[] shapes ) { for( int i=0; i < shapes.length; ++i ) { shape[i].draw( device ); }

16 ISP – Interface Segregation Principle Clients should not be forced to depend upon services they do not use. - R.Martin, 1996 Hints:  Avoid fat interfaces  Separate clients mean separate interfaces Violation cost:  lack of flexibility

17 ISP violation: Security Door Door has to sound an alarm if it is open for too long. Problem: Timeout method has to be public. But…there are Door clients that do NOT use timeout method and don’t have to. This approach leads to mistakes.

18 ISP-compliant Security Door Door clients still can use TimedDoor via Door interface Door clients will not be affected by changes made in Timer, TimerClient and TimedDoor

19 ISP violation: ATM Transactions  Adding new transactions causes all other transactions to recompile  If any Transaction requires a change to UI, all of the other will be forced to recompile

20 ISP-compliant solution

21 Dependencies  MDPMinimal Dependencies Principle  DIPDependency Inversion Principle  ADPAcyclic Dependencies Principle

22 DIP – Dependency Inversion Principle High level modules should not depend upon low-level ones. Both should depend upon abstractions. Abstractions should not depend upon implementation details. Details should depend upon abstractions. - R.Martin, 1996

23 DIP violation: Copier What if we need to support another kind of printer ?

24 DIP-compliant solution Now we can easily add new writers and readers

25 ADP – Acyclic Dependencies Principle The dependency structure between entities (classes, packages, functions) must be a Directed Acyclic Graph (DAG). - R.Martin, 1996  Two entities having to know about each other can not be used separately. They work like a monolith and there is no benefit in separating them.  Increases maintainability

26 Example: cyclic dependencies Due to the dependency from MyDialogs to Application, MyTasks package depends upon the entire system.

27 Package design principles  REP - Reuse-Release Equivalence Principle  CRP - Common Reuse Principle  CCP - Common Closure Principle  SDP - Stable Dependencies Principle  SAP - Stable Abstractions Principle

28 REP - Reuse/Release Equivalence Principle The unit of reuse is the unit of release. The unit is what UML refers to as a package. - R.Martin, 1996  Classes should be grouped into packages according to how they will be reused  To be effectively reused, packages must be given a release number  Makes updates convenient for reusers

29 CRP – Common Reuse Principle The classes in a package are to be reused together. Reusers should depend upon the entire package, not just a part of it. - R.Martin, 1996  ISP, scaled to packages  Reduces maintenance cost, increases reusability

30 CRP violation: remote service Problem: Every time the new version of Service is released, clients of ServiceAgent must expect that their code won’t work, even if changes do not affect ServiceAgent.

31 CRP compliant design Clients of ServiceAgent depend only upon things they really use. Benefit: Application can easily switch from local to remote service implementation.

32 CRP compliant design Clients of ServiceAgent depend only upon things they really use. Benefit: Changes in local and server packages do not affect application

33 CCP – Common Closure Principle Classes within the package should be affected by the same kind of changes. Either all open to the kind of change or all closed to the kind of change. - R.Martin, 1996  This produces packages where changes are very localized, and, therefore, number of releases is minimized.

34 ADP – Acyclic Dependencies Principle The dependency structure between packages must be a Directed Acyclic Graph (DAG). - R.Martin, 1996  Two packages having to know about each other can’t be reused separately. Therefore, they work like a single package.  Increases maintainability

35 SDP – Stable Dependencies Principle A package should only depend upon packages that are more stable than itself. Stability is a measure of difficulty in changing a package. - R.Martin, 1996  What makes a package hard to change?  Increases maintainability

36 Measuring Instability Instability of a package I = Ce / ( Ca + Ce ) where Ce = efferent couplings (number of classes outside the package that classes inside the package depend upon). How dependent am I? Ca = afferent couplings (number of classes outside the package that depend upon classes within the package). How responsible am I?

37 SAP – Stable Abstractions Principle The abstractness of the package should be proportional to its stability. - R.Martin, 1996  If all packages are maximally stable, the system would be unchangeable.  Therefore, some packages must be instable.  Increases maintainability

38 Measuring Abstractness Abstractness of a package A = Na / N where Na = number of abstract classes N = total number of classes

39 Main sequence Main Sequence – an instability versus abstractness graph Packages along the line from (0,1) to (1,0) have a good balance Distance from the main sequence D = | A + I – 1 | Given this metric, a design can be analyzed for its overall conformance to the main sequence. - R.Martin, 1994

40 Main sequence I A 1 1 0 USELESS AREA CONCRETE CLASSES

41 Example: server  What happens if ResultSet is not an interface? I=0 A=1 D=0 I = 1/1 = 1 A= 0 D= 0 I=2/3 A=0 D=1/3

42 Example: server I=1 A=0 D=0 I=0 A=1/2 D=1/2 I=1/2 A=0 D=1/2 This change affects both MyServer and AbstractServer balance


Download ppt "OO Design Principles Copyright © Vyacheslav Mukhortov, Nikita Nyanchuk-Tatarskiy, 2001-2004 Copyright © INTEKS LLC, 2003-2004."

Similar presentations


Ads by Google