Comparison of Object-Oriented and Paradigm Independent Software Complexity Metrics

Comparison of Object-Comparison of Object-Oriented andOriented and

Paradigm IndependentParadigm IndependentSoftware Complexity MetricsSoftware Complexity Metrics

ZolZoltán Porkoláb, Ádám Sillyetán Porkoláb, Ádám SillyeDepartment of Programming Languages and CompilersDepartment of Programming Languages and Compilers

Eötvös Loránd University, Faculty of InformaticsEötvös Loránd University, Faculty of Informatics

e-mail: {ge-mail: {gsdsd|madic}@elte.hu|madic}@elte.hu

22ICAI 6 Eger - Comparison of Object-Oriented and Paradigm Independent Software Complexity MetricsICAI 6 Eger - Comparison of Object-Oriented and Paradigm Independent Software Complexity Metrics

The structure of the presentationThe structure of the presentation

The role of software metricsThe role of software metrics

Metrics: an overviewMetrics: an overview

Object-oriented software metricsObject-oriented software metrics

Multiparadigm programmingMultiparadigm programming

The AV-graphThe AV-graph

Empirical resultsEmpirical results


Role of software metricsRole of software metrics

Cost of the softwareCost of the software SpecificationSpecification DesignDesign ImplemeImplementationntation TesTesting & Bug-fixingting & Bug-fixing MaintenanceMaintenance

More than More than 70% 70% of cost is for testing and maintenanceof cost is for testing and maintenance (Zuse 1998)(Zuse 1998)

Software qualitySoftware quality


SoftSoftwareware metricsmetrics

Software metricsSoftware metrics:: Measuring the development process (process metrics)Measuring the development process (process metrics) Measuring the productMeasuring the product ( (product metricsproduct metrics))

Product metricsProduct metrics:: External metricsExternal metrics::

Reliability metricsReliability metricsFunctional metricsFunctional metricsEfficiency metricsEfficiency metrics

Internal product metricsInternal product metrics::SizeSize ComplexityComplexityStStyleyle


Product metricsProduct metrics

Size metricsSize metrics:: LOC, eLOCLOC, eLOC Ignore the semanticIgnore the semantic

Structural metricsStructural metrics::

McCabe 1976McCabe 1976 MotivationMotivation: : predict testing effortspredict testing efforts For structured programsFor structured programs: V(G) = p + 1: V(G) = p + 1

HowattHowatt and and Baker 1989 Baker 1989 MotivationMotivation: : involve nesting levelinvolve nesting level SN(G) = |N|+ND(G)SN(G) = |N|+ND(G)


Object-oriented metrics in theoryObject-oriented metrics in theory

Chidamber-Kemerer (1994)Chidamber-Kemerer (1994) OO metrics suite: OO metrics suite: WMC (Weighted Methods per Class)WMC (Weighted Methods per Class) DIT (Depth of Inheritance Tree, DIT)DIT (Depth of Inheritance Tree, DIT) NOC (Number of ChildNOC (Number of Child Classes Classes)) CBO (Coupling Between Object Classes)CBO (Coupling Between Object Classes) fan-in fan-in

and fan outand fan out RFC (Response for Class)RFC (Response for Class) LCOM (Lack of Cohesion in Methods)LCOM (Lack of Cohesion in Methods)

Chidamber - KemererChidamber - Kemerer

Henderson - SellersHenderson - Sellers


Object-Oriented metrics in practiceObject-Oriented metrics in practice

EclipseEclipse ( (www.eclipse.orgwww.eclipse.org))

McCabe McCabe cyclomatic complexitycyclomatic complexity Efferent couplingEfferent coupling ( (≠ fan out!)≠ fan out!) Lack of CohesionLack of Cohesion LOC (Lines of Code)LOC (Lines of Code) Number of Fields in ClassesNumber of Fields in Classes Nesting DepthNesting Depth Number of Number of Method Method ParametersParameters Number of StatementsNumber of Statements Weighted Methods per ClassWeighted Methods per Class


Software paradigmSoftware paradigm

Evolution of Software paradigmEvolution of Software paradigm Structured programmingStructured programming Object-oriented programmingObject-oriented programming

Classes, inheritance, virtual functionClasses, inheritance, virtual function Generative programming Generative programming

Aspect-Oriented (Kitzales 1994)Aspect-Oriented (Kitzales 1994)

Intentional (Simonyi 1995)Intentional (Simonyi 1995)

Template metaprogramming (Veldhuizen 1994)Template metaprogramming (Veldhuizen 1994)

Multiparadigm programming (Coplien 1998)Multiparadigm programming (Coplien 1998) Simultaneous usage of paradigmsSimultaneous usage of paradigms


Paradigm-independent Software MetricsParadigm-independent Software Metrics

Applicable for programs written in different Applicable for programs written in different paradigms or paradigms or in in mixed-paradigm environment mixed-paradigm environment

Based on general programming language features which Based on general programming language features which are paradigm- and language-independent.are paradigm- and language-independent.The paradigm-dependent attributes are derived fromThe paradigm-dependent attributes are derived fromthese features.these features.


ComponentsComponents

Control Structure of ProgramControl Structure of Program Most of the programs share the same control Most of the programs share the same control

statementsstatements

Complexity of Data TypesComplexity of Data Types Reflects the complexity of data types used (like Reflects the complexity of data types used (like

classes)classes)

Complexity of Data AccessComplexity of Data Access Connection between control structure and data Connection between control structure and data

Direction of data flowDirection of data flowNesting depthNesting depth


AV-graphAV-graph

smain

P1

tmain

b

P4

ac

d1

d3d4

d2

data node

output node

input node


AV-grAV-graphaph metrics metrics

|N||N| number of nodesnumber of nodes

nd(n) = | Pred(n) |nd(n) = | Pred(n) | nesting depth of node ‘n’ nesting depth of node ‘n’

ND(G) = ND(G) = nd(n)nd(n) for all nodes for all nodes

C(G) = |N’| + ND(G)C(G) = |N’| + ND(G)

C(O) = |N’| + C(O) = |N’| + ND(G) ND(G)

= |A| + = |A| + ( (ND(G) + |L|)ND(G) + |L|)


Complexity of classComplexity of class

Control structureControl structuress Complexity of Complexity of method method control structurecontrol structuress

Complexity of data typeComplexity of data typess Local variables in methodsLocal variables in methods Attributes (could be complex types)Attributes (could be complex types) Coupling between classesCoupling between classes InheritanceInheritance

Complexity of data handlingComplexity of data handling Connection between control structure and dataConnection between control structure and data


class date{public: void set_next_month() { if ( month == 12 ) { month = 1; year = year + 1; } else { month = month + 1; } } void set_next_day() { if ( month == 1 || month == 3 || ... || month == 12 ) if ( day == 31 ) set_next_month(); else day = day + 1; else if ( day == 30 ) set_next_month(); else day = day + 1; }private:

int year, month, day;};

Complexity of ClassComplexity of Class

sset_next_month

P1

tsnm

b

a c

sset_next_day

P2

tsnm

e

P3

gf

P4d1

d3

d2


Our Our Measuring toolMeasuring tool

Supported lSupported languageanguagess Java 1.3Java 1.3 Java 1.4Java 1.4 (assert) (assert) Future directionFuture directionss: C# and C++: C# and C++

ImplementationImplementation ANTLRANTLR User interface: sUser interface: standalonetandalone application and application and

Eclipse Plug-inEclipse Plug-in

OuputOuput CSVCSV XMLXML


Applied MetricsApplied Metrics

Object-Oriented MetricsObject-Oriented Metrics Inner Class DepthInner Class Depth Inheritance levelInheritance level Number of ChildrenNumber of Children Number of MethodsNumber of Methods Number of FieldsNumber of Fields LCOMLCOM Henderson-SellersHenderson-Sellers LCOM* LCOM* Fan-outFan-out


Applied MetricsApplied Metrics

Size MetricsSize Metrics eLOCeLOC Number of StatementsNumber of Statements

McCabeMcCabe

Howatt-BakerHowatt-Baker

AV-graphAV-graph


Test DataTest Data

Java Standard Library 1.4.2Java Standard Library 1.4.2 367367.000.000 eLOC eLOC

jBOSS 3.2.3jBOSS 3.2.3 300300.000.000 eLOC eLOC

Omg.org.CORBAOmg.org.CORBA 55.000.000 eLOC eLOC

The measure tool (with mostly generated parser)The measure tool (with mostly generated parser) 77.000.000 eLOC eLOC

Eclipse 3.0M6Eclipse 3.0M6 900900.000.000 eLOCeLOC

17.000 17.000 cclasslass – more than 1.5 million lines – more than 1.5 million lines


Calibrating our toolCalibrating our tool

Goal: Goal: Achieve similar results with other measuring Achieve similar results with other measuring

programs. (Otherwise we are unable to compare the programs. (Otherwise we are unable to compare the results.)results.)

DifficultiesDifficulties Metrics are not defined precisely enough from the Metrics are not defined precisely enough from the

view of implementation.view of implementation. Available toolsAvailable tools

working on different abstract program tree.working on different abstract program tree.use simplified metrics definitions.use simplified metrics definitions.and have several bugs.and have several bugs.


ExampleExample

Lack of Cohesion in MethodsLack of Cohesion in Methods P = method pairs that access different fieldsP = method pairs that access different fields Q = method pairs that access at least one common Q = method pairs that access at least one common

fieldfield LCOM = |P|-|Q| if |P|-|Q|LCOM = |P|-|Q| if |P|-|Q|≥0, otherwise 0.≥0, otherwise 0.

References to References to thisthis and and supersuper have to be counted or have to be counted or not in the sense of cohesion?not in the sense of cohesion? No, but some tools count them.No, but some tools count them.


ResultsResults

No statistical correlation between the OO and multi No statistical correlation between the OO and multi paradigm metrics:paradigm metrics: OO metrics only measure the big pictureOO metrics only measure the big picture MPM consider more properties: higher densityMPM consider more properties: higher density The structural complexity of methods are extremely The structural complexity of methods are extremely

increase the overall complexity.increase the overall complexity.


ConclusionConclusion

Implementations are not reliable yet:Implementations are not reliable yet: No precise definition for metricsNo precise definition for metrics Different input data: languagesDifferent input data: languages Tools:Tools:

Various interpretation of definitions.Various interpretation of definitions.

Various bugs.Various bugs.

Results of different tools cannot be compared.Results of different tools cannot be compared.

Future examination...Future examination...

Comparison of Object-Oriented and Paradigm Independent Software Complexity Metrics

Documents

Transcript of Comparison of Object-Oriented and Paradigm Independent Software Complexity Metrics