Finding Analysis Classes-Object-Oriented Design-Lecture 7 Slides-Computer Engineering, Slides of Object Oriented Analysis and Design

Download Finding Analysis Classes-Object-Oriented Design-Lecture 7 Slides-Computer Engineering and more Slides Object Oriented Analysis and Design in PDF only on Docsity!

Department of Computer Engineering

Sharif University of Technology

Object-Oriented Design

Lecturer: Raman Ramsin

Lecture 7 : Finding Analysis Classes

Department of Computer Engineering

Sharif University of Technology

Analysis Workflow: Analyze a Use Case  The analysis workflow consists of the following activities:  Architectural analysis  Analyze a use case  Outputs:  analysis classes  use case realizations  Analyze a class  Analyze a package

Department of Computer Engineering

Sharif University of Technology

Good Analysis Classes  What makes a good analysis class?  Its name reflects its intent.  It is a crisp abstraction that models one specific element of the problem domain.  It maps to a clearly identifiable feature of the problem domain.  It has a small, well-defined set of responsibilities:  a responsibility is a contract or obligation that a class has to its clients;  a responsibility is a semantically cohesive set of operations;  there should only be about three to five responsibilities per class.  It has high cohesion - all features of the class should help to realize its intent.  It has low coupling - a class should only collaborate with a small number of other classes to realize its intent.

Department of Computer Engineering

Sharif University of Technology

Bad Analysis Classes  What makes a bad analysis class?  A functoid - a class with only one operation.  A stand-alone class - each class should be associated with a small number of other classes with which it collaborates to deliver the desired benefit.  An omnipotent class - a class that does everything (classes with "system" or "controller" in their name may need closer scrutiny).  A class with a deep inheritance tree - in the real world inheritance trees tend to be shallow.  A class with low cohesion.  A class with high coupling.  Many very small classes in a model – merging should be considered.  Few but large classes in a model – decomposition should be considered.

Department of Computer Engineering

Sharif University of Technology

Noun/verb analysis ( Grammatical Parsing)

1. Collect as much relevant information about the problem domain as possible; suitable sources of information are:  The requirements model  The use case model  The project glossary  Any other document (architecture, vision documents, etc.)
2. Analyze the documentation:  Look for nouns or noun phrases - these are candidate classes or attributes.  Look for verbs or verb phrases - these are candidate responsibilities or operations.
3. Make a tentative allocation of the attributes and responsibilities to the classes.

Department of Computer Engineering

Sharif University of Technology

CRC Analysis – CRC Cards  CRC – Class, Responsibilities, and Collaborators  Important things in the problem domain are written on CRC Cards. Each Card has three compartments:  class - contains the name of the class  responsibilities - contains a list of the responsibilities of that class (the functions it performs and even the information it is responsible to keep and provide)  collaborators - contains a list of other classes with which this class collaborates in order to fulfill the responsibilities

Department of Computer Engineering

Sharif University of Technology

CRC Analysis Procedure – Phase 2  The participants are 00 analysts and domain experts.  Phase 2 : Decide which sticky notes should become classes and which should become attributes:  Analysis classes must represent a crisp abstraction in the problem domain. Certain sticky notes will represent key business concepts and clearly need to become classes.  If a note logically seems to be a part of another note, this is a good indication that it represents an attribute.  If a note doesn't seem to be particularly important or has very little interesting behavior, see if it can be made an attribute of another class.  If in doubt about a note, just make it a class.

Department of Computer Engineering

Sharif University of Technology

Use-Case-Based Analysis  Complements other techniques  Starts from an initial list of classes  List of classes is perfected and refined based on use cases:  Behavioral models are built showing use case realizations  Classes are identified based on the objects needed for use case realizations: the list of classes should provide instances which implement the behavior needed for the use cases;

 New classes will be added if needed

 Changes will be made to existing classes if required for use case realization

Department of Computer Engineering

Sharif University of Technology

Real-World Analysis  Explore the real world for classes:  Candidates: physical objects, paperwork, interfaces to the outside world, and conceptual entities;  Physical objects: Things such as aircraft, people, and hotels may all indicate classes.  Paperwork: Things like invoices, orders, and bankbooks may all indicate possible classes; beware of paperwork supporting the redundant business processes that the new system might be trying to replace.  Known interfaces to the outside world: Things such as screens, keyboards, peripherals, and other systems can be a source of candidate classes, especially for embedded systems.  Conceptual entities: Things that are crucial to the operation of the business but are not manifest as concrete things; such as enrollment, educational program, and alarm condition.

Department of Computer Engineering

Sharif University of Technology

Analysis Model  Create a first-cut analysis model:  compare noun/verb analysis results with CRC results and the results of an examination of other sources of classes.  resolve synonyms and homonyms.  differences between the results of the different techniques indicate areas of uncertainty.  consolidate results into a first-cut analysis model.