第五章集成測試

第5章集成測試5.1深入研究SATM系統(tǒng)5.2集成測試的主要方法5.2.1基于功能分解的集成5.2.2基于調(diào)用圖的集成5.2.3基于路徑的集成5.3案例研究1西北工業(yè)大學(xué)軟件與微電子學(xué)院SATM系統(tǒng)的功能分解1A10BDE111213141523456789C1617F22181920212324252627Table1:SATMUnitsandAbbreviatedNamesUnitLevelUnitName11SATMSystemA1.1DeviceSense&ControlD1.1.1DoorSense&Control21.1.1.1GetDoorStatus31.1.1.2ControlDoor41.1.1.3DispenseCashE1.1.2SlotSense&Control51.1.2.1WatchCardSlot

61.1.2.2GetDepositSlotStatus71.1.2.3ControlCardRoller81.1.2.3ControlEnvelopeRoller91.1.2.5ReadCardStrip101.2CentralBankComm.111.2.1GetPINforPAN121.2.2GetAccountStatus131.2.3PostDailyTransactions

UnitLevelUnitNamB1.3TerminalSense&Control141.3.1ScreenDriver151.3.2KeySensorC1.4ManageSession161.4.1ValidateCard171.4.2ValidatePIN181.4.2.1

GetPIN

F1.4.3CloseSession191.4.3.1NewTransactionRequest201.4.3.2PrintReceipt211.4.3.3PostTransactionLocal221.4.4ManageTransaction231.4.4.1GetTransactionType241.4.4.2

GetAccountType251.4.4.3

ReportBalance261.4.4.4

ProcessDeposit271.4.4.5

ProcessWithdrawal2西北工業(yè)大學(xué)軟件與微電子學(xué)院5.2.1基于功能性分解的集成兩種表示方式樹和文字形式集成的4種模塊順序從樹頂開始向下：自頂向下(Top-down)從樹底開始向上：自底向上(Bottom-up)三明治方法:Sandwich大爆炸方法：BigBang3西北工業(yè)大學(xué)軟件與微電子學(xué)院大爆炸集成測試UnitTestHUnitTestDUnitTestCUnitTestBUnitTestASystem-WideTest………….Allcomponentsareassembledandtestedatonce.4西北工業(yè)大學(xué)軟件與微電子學(xué)院自頂向下集成TopSubtree

(Sessions1-4)SecondLevelSubtree

(Sessions12-15)BotomLevelSubtree

(Sessions38-42)5西北工業(yè)大學(xué)軟件與微電子學(xué)院自頂向下集成（續(xù)）6西北工業(yè)大學(xué)軟件與微電子學(xué)院AdvantagesandDisadvantagesofTop-DownIntegrationTestingTestcasescanbedefinedintermsofthefunctionalityofthesystem(functionalrequirements).StructuraltechniquescanalsobeusedfortheunitsinthetoplevelsWritingstubscanbedifficultespeciallywhenparameterpassingiscomplex.StubsmustallowallpossibleconditionstobetestedPossiblyaverylargenumberofstubsmayberequired,especiallyifthelowestlevelofthesystemcontainsmanyfunctionalunitsOnesolutiontoavoidtoomanystubs:Modifiedtop-downtestingstrategy(Bruege)TesteachlayerofthesystemdecompositionindividuallybeforemergingthelayersDisadvantageofmodifiedtop-downtesting:Both,stubsanddriversareneeded7西北工業(yè)大學(xué)軟件與微電子學(xué)院自底向上集成TopSubtree(Sessions29-32)SecondLevelSubtree(Sessions25-28)BottomLevelSubtree(Sessions13-17)8西北工業(yè)大學(xué)軟件與微電子學(xué)院自底向上集成（續(xù)）9西北工業(yè)大學(xué)軟件與微電子學(xué)院AdvantagesandDisadvantagesofBottom-UpIntegrationTestingNotoptimalstrategyforfunctionallydecomposedsystems:Teststhemostimportantsubsystem(UI)lastUsefulforintegratingthefollowingsystemsObject-orientedsystemsReal-timesystemsSystemswithstrictperformancerequirements10西北工業(yè)大學(xué)軟件與微電子學(xué)院三明治集成11西北工業(yè)大學(xué)軟件與微電子學(xué)院12西北工業(yè)大學(xué)軟件與微電子學(xué)院AdvantagesandDisadvantagesofSandwichIntegrationTestingTopandBottomLayerTestscanbedoneinparallelDoesnottesttheindividualsubsystemsthoroughlybeforeintegrationSolution:Modifiedsandwichtestingstrategy(Bruege)13西北工業(yè)大學(xué)軟件與微電子學(xué)院StepsinIntegration-Testing

.1.Basedontheintegrationstrategy,selectacomponenttobetested.Unittestalltheclassesinthecomponent.2.Putselectedcomponenttogether;doanypreliminaryfix-upnecessarytomaketheintegrationtestoperational(drivers,stubs)3.Dofunctionaltesting:Definetestcasesthatexerciseallusescaseswiththeselectedcomponent4.Dostructuraltesting:Definetestcasesthatexercisetheselectedcomponent5.Executeperformancetests6.Keeprecordsofthetestcasesandtestingactivities.7.Repeatsteps1to7untilthefullsystemistested.Theprimarygoalofintegrationtestingistoidentifyerrorsinthe(current)componentconfiguration.14西北工業(yè)大學(xué)軟件與微電子學(xué)院WhichIntegrationStrategyshouldyouuse?FactorstoconsiderAmountoftestharness(stubs&drivers)LocationofcriticalpartsinthesystemAvailabilityofhardwareAvailabilityofcomponentsSchedulingconcernsBottomupapproachgoodforobjectorienteddesignmethodologiesTestdriverinterfacesmustmatchcomponentinterfaces......Top-levelcomponentsareusuallyimportantandcannotbeneglecteduptotheendoftestingDetectionofdesignerrorspostponeduntilendoftestingTopdownapproachTestcasescanbedefinedintermsoffunctionsexaminedNeedtomaintaincorrectnessofteststubsWritingstubscanbedifficultRef:“Object-OrientedSoftwareEngineering–UsingUML,PatternsandJava”,BerndBruegge,AllenH.Dutoit15西北工業(yè)大學(xué)軟件與微電子學(xué)院優(yōu)缺點優(yōu)點直覺上清晰容易根據(jù)分解樹跟蹤缺點人工開發(fā)、樁或驅(qū)動器的開發(fā)工作量是這些方法的缺點、還有重新測試所需的工作量問題給定分解樹所需集成測試會話數(shù)的計算公式（一個測試過程是為一個特定配置的實際代碼和樁所做的一組測試）會話=節(jié)點-頁+邊SATM系統(tǒng)有42個集成測試過程-》42個單獨(dú)的集成測試用例集合對于自頂向下集成，需開發(fā)(節(jié)點-1個)樁；對于自底向上集成，需開發(fā)(節(jié)點-葉個)驅(qū)動器。對于SATM系統(tǒng)，為32個樁和10個驅(qū)動器16西北工業(yè)大學(xué)軟件與微電子學(xué)院基于調(diào)用圖的集成成對集成思想是免除樁/驅(qū)動器開發(fā)工作相鄰集成可大大降低集成測試會話數(shù)量，且避免了樁和驅(qū)動器的開發(fā)計算公式內(nèi)部節(jié)點=節(jié)點-（源節(jié)點+匯節(jié)點）鄰居=內(nèi)部節(jié)點+源節(jié)點鄰居=節(jié)點–匯節(jié)點17西北工業(yè)大學(xué)軟件與微電子學(xué)院18西北工業(yè)大學(xué)軟件與微電子學(xué)院19西北工業(yè)大學(xué)軟件與微電子學(xué)院20西北工業(yè)大學(xué)軟件與微電子學(xué)院基于路徑的集成21西北工業(yè)大學(xué)軟件與微電子學(xué)院22西北工業(yè)大學(xué)軟件與微電子學(xué)院23西北工業(yè)大學(xué)軟件與微電子學(xué)院Inaddition,thefollowingaremoduleexecutionpaths:MEP(A,1)=<1,2,3,6> MEP(B,1)=<1,2>MEP(A,2)=<1,2,4> MEP(B,2)=<3,4>MEP(A,3)=<5,6> MEP(C,1)=<1,2,4,5>MEP(C,2)=<1,3,4,5>

ExecutionPathsExample

24西北工業(yè)大學(xué)軟件與微電子學(xué)院Givenasetofunits,theirMM-Pathgraphisthedirectedgraphinwhichnodesaremoduleexecutionpathsand,edgescorrespondtomessagesandreturnsfromoneunittoanotherThedefinitioniswithrespecttoasetofunits.Itdirectlysupportscompositionofunitsandcompositionbasedintegrationtesting.Itispossibletocomposedowntolevelofindividualmoduleexecutionpathsbutitwouldbemoredetailedthannecessary.

MM-PathGraph

25西北工業(yè)大學(xué)軟件與微電子學(xué)院MEP(A,2)MEP(B,1)MEP(C,1)MEP(B,2)MEP(A,3)MEP(A,1)MEP(C,2)Figure2

TheFigure2aboveillustratestheMM-PathgraphfortheexampleinFigure1.Thesolidarrowsindicatemessagesandthecorrespondingreturnsareindicatedbydottedarrows.

MM-PathGraphExample

Ref:“SoftwareTestingACraftsman'sApproach”2ndedition,PaulC.Jorgensen26西北工業(yè)大學(xué)軟件與微電子學(xué)院MM-Pathsimplementafunctionthattranscendsunitboundaries,thuswehavethefollowingrelationship:Considerthe“intersection”ofanMM-Pathwithaunit.Themoduleexecutionpathsinthisintersectionareananalogofaslicewithrespecttothe(MM-Path)function.Hencethemoduleexecutionpathsinsuchanintersectionaretherestrictionofthefunctiontotheunitinwhichtheyoccur.

MM-PathAnalogy

27西北工業(yè)大學(xué)軟件與微電子學(xué)院TherearethreeobservablebehavioralcriteriathatputendpointsonMM-Paths:eventquiescence:occurswhenasystemisnearlyidle,waitingforaportinputeventtotriggerfurtherprocessing.Thisisasystemlevelpropertywithananalogattheintegrationlevel:messagequiescence.messagequiescence:occurswhenaunit