Skip to main content
UploadLanguage (EN)Support
BusinessMobileSocial MediaMarketingTechnologyArt & PhotosCareerDesignEducationPresentations & Public SpeakingGovernment & NonprofitHealthcareInternetLawLeadership & ManagementAutomotiveEngineeringSoftwareRecruiting & HRRetailSalesServicesScienceSmall Business & EntrepreneurshipFoodEnvironmentEconomy & FinanceData & AnalyticsInvestor RelationsSportsSpiritualNews & PoliticsTravelSelf ImprovementReal EstateEntertainment & HumorHealth & MedicineDevices & HardwareLifestyle
Ptidej Team, profile picture
Uploaded byPtidej Team
160 views

CSMR08.ppt

This study examines whether design patterns as presented in the Gang of Four book positively impact software quality characteristics. An empirical study was conducted where developers rated the impact of 23 design patterns on 10 quality attributes. Results showed that overall patterns had a positive impact on characteristics like expandability and reusability. However, impact varied by pattern and quality attribute, with some negative impacts observed. The study thus concludes that while design patterns can improve quality, their impact is mixed and they should be applied carefully.

Technology◦Design◦
Download to read offline
Do Design Patterns Impact Software Quality Positively? Foutse Khomh and Yann-Gaël Guéhéneuc foutsekh@iro.umontreal.ca, guehene@iro.umontreal.ca CSMR'08 GEODES Ptidej Team – OO Programs Quality Evaluation and Enhancement using Patterns Group of Open, Distributed Systems, Experimental Software Engineering Department of Informatics and Operations Research University of Montreal © Khomh, 2008
Context Design patterns (as presented by the GoF) have been said to have a positive impact on quality characteristics – Expandability – Reusability – Understandability But, some studies presented counter-examples, i.e., Wendorff (Assessment of design patterns during software reengineering: Lessons learned from a large commercial project.) 2/??
Study Design (1/2) Hypothesis: “Design patterns have a positive impact on quality characteristics” Objective: prove or disprove the hypothesis with concrete evidence 3/??
Study Design (2/2) Required choices – Quality characteristics – Kind of study Our choices – 10 quality characteristics related to design, implementation, and runtime – Empirical study (vs. absolute, vs. relative) • Manual activity • Developers' perception 4/??
Method Choice of design patterns of interest – 23 design patterns in the GoF book Identification and definition of quality characteristics of interest – 10 characteristics Design Implementation Runtime Expandability Learnability Generality Simplicity Understandability Modularity at runtime Reusability Modularity Scalability Robustness Definition of a scale 5/?? – Likert scale of 5 points + "I don't know"
Data Collection Questionnaire send out to colleagues with a good knowledge of design patterns between January and May 2007 Questionnaire posted on mailing-lists refactoring@yahoogroups.com patterns-discussion@cs.uiuc.edu 6/?? gang-of-4-patterns@cs.uiuc.edu
Data Processing Number of received questionnaires: 30 – Number of complete questionnaires: 18 – Number of almost complete (90%) questionnaires: 2 – Incomplete questionnaires or with “I don't know”:10 Number of kept questionnaires: 20 Answers A and B aggregated in Positive, C in Neutral, and D and E in Negative – Some questionnaire were clearly generally more positive/negative than others 7/??
Results (1/2) 8/??
Results (2/2) We discuss detailed results – Three well-known design patterns • Composite • Abstract Factory • Flyweight – Three quality characteristics mentioned in the GoF • Reusability • Expandability 9/?? • Understandability
Composite Overall positive impact on quality attributes Scalability and robustness not so positively impacted 10/??
Abstract Factory Overall neutral impact on quality attributes Expandability, generality, and modularity as expected Robustness surprising 11/??
Flyweight Overall negative impact on quality attributes Positive impact only on scalability as expected 12/??
Reusability Positive Neutral Negative A.Factory Builder Bridge 10 positive impacts F.Method Singleton Decorator Prototype Facade Flyweight 6 neutral impacts Adapter Mediator Command Composite Ch.Of.Resp. Memento Proxy Interpreter State 7 negative impacts Iterator Visitor Observer Strategy T.Method 13/??
Expandability Positive Neutral Negative A.Factory Flyweight Singleton 19 positive impacts Builder Memento F.Method Proxy 3 neutral impacts Prototype Adapter Bridge 1 negative impact Composite Decorator Facade Ch.Of.Resp. Command Interpreter Iterator Mediator Observer State Strategy T.Method Visitor 14/??
Understandability Positive Neutral Negative A.Factory Proxy Decorator 15 positive impacts Builder Memento Flyweight F.Method T.Method Ch.Of.Resp. 3 neutral impacts Prototype Command Singleton Adapter Visitor 5 negative impacts Bridge Composite Facade Interpreter Iterator Mediator Observer State Strategy 15/??
Null Hypothesis Test 16/??
Conclusion (1/2) Globally, design patterns have a positive impact on quality attributes Quality Expandability Simplicity Reusabilit Learnabilit Understandabilit Characteristics y y y Number of 19+ / 4- 16+ / 7- 11+ / 12- 14+ / 9- 11+ / 12- Positive/Negative Patterns Quality Characteristics Modularity Generality Mod. at runtime Scalability Robustness Number of 18+ / 5- 19+ / 4- 11+/ 12- 1+ / 22- 0+ / 23- Positive/Negative Patterns 17/??
Conclusion (2/2) Pattern by pattern, their impact on quality varies and includes negative impact! So, design patterns do not always improve the quality of systems, thus they should be use with caution 18/??
Help Wanted Please help us refine this study by downloading and filling in the questionnaire http://www.ptidej.net/downloads/ 19/??