ModernSoftwareEngineering蔣欣榮1軟件工程學(xué)科介紹概念提出1968年NATO會議(北大西洋公約組織的計(jì)算機(jī)科學(xué)家的國際會議〕提出“軟件工程〞概念。當(dāng)時對“軟件工程〞代表性定義：為了經(jīng)濟(jì)地獲得在真實(shí)機(jī)器上可靠工作的軟件而制定和使用的合理工程原那么和方法。1972年IEEE學(xué)會的計(jì)算機(jī)分會IEEE—CS第一次出版了“軟件工程學(xué)報(bào)〞2軟件工程學(xué)科介紹學(xué)科雛形上世紀(jì)70年代末，美國將軟件工程教程列入研究生教學(xué)方案。1980年代末和1990年代初，軟件工程教育得到卡內(nèi)基-梅隆大學(xué)軟件工程研究所〔CMU/SEI〕的支持。1991年，“軟件工程〞被ACM〔美國計(jì)算機(jī)協(xié)會〕和IEEE-CS列為計(jì)算科學(xué)的九個知識領(lǐng)域之一。1993年，IEEE-CS和ACM為了把軟件工程建設(shè)成為一個專業(yè)，建立了IEEE-CS/ACM聯(lián)合指導(dǎo)委員會。3軟件工程學(xué)科介紹學(xué)科確立2004年6月，IEEE-CS和ACM聯(lián)合給出：SWEBOK(軟件工程知識體)〔SoftwareEngineeringBodyofKnowledge)把軟件工程學(xué)科的主體知識分為10個知識領(lǐng)域：軟件需求 軟件設(shè)計(jì) 軟件構(gòu)造軟件測試 軟件維護(hù) 軟件配置管理軟件工程管理 軟件工程過程 軟件質(zhì)量軟件工程工具和方法 4軟件工程與其他學(xué)科的關(guān)系軟件工程是計(jì)算科學(xué)9個領(lǐng)域之一算法和數(shù)據(jù)結(jié)構(gòu)程序設(shè)計(jì)語言計(jì)算機(jī)系統(tǒng)結(jié)構(gòu)數(shù)值和符號計(jì)算操作系統(tǒng)軟件方法學(xué)和軟件工程數(shù)據(jù)庫和信息檢索人工智能和機(jī)器人學(xué)人機(jī)交互5軟件工程是一門工程學(xué)科是指導(dǎo)計(jì)算機(jī)軟件開發(fā)與維護(hù)的一門工程學(xué)科。工程：是科學(xué)和數(shù)學(xué)的某種應(yīng)用，通過這一應(yīng)用，使自然界的物質(zhì)和能源的特性能夠通過各種結(jié)構(gòu)、機(jī)器、產(chǎn)品、系統(tǒng)和過程，以最短的時間和精而少的人力做出高效、可靠且對人類有用的東西。將自然科學(xué)的理論應(yīng)用到具體工農(nóng)業(yè)生產(chǎn)部門中形成的各學(xué)科的總稱。包括：研究、開發(fā)、設(shè)計(jì)、施工、生產(chǎn)、操作、管理及其他職能等。典型的傳統(tǒng)工程：建筑工程、機(jī)械工程、電力工程等。軟件工程即用工程、科學(xué)和數(shù)學(xué)的原那么和方法研制、維護(hù)計(jì)算機(jī)軟件的有關(guān)技術(shù)及方法，其優(yōu)點(diǎn)是以較小的代價(jià)開發(fā)高質(zhì)量的軟件并有效地維護(hù)它。6簡單的建筑

修建簡單的建筑僅需要初級的建筑知識，如建立房屋需要立柱、橫梁、屋頂及墻面等根本元素；在頭腦中大致構(gòu)思，便可以建造。7復(fù)雜的建筑

復(fù)雜的建筑是許多人合作的結(jié)果，不可能一邊構(gòu)思，一邊修建。在實(shí)際開工前，必須完成全部〔至少是絕大多數(shù)〕設(shè)計(jì)工作，并經(jīng)過評審，才能開工。其設(shè)計(jì)包括外觀設(shè)計(jì)、平面設(shè)計(jì)、結(jié)構(gòu)設(shè)計(jì)、施工設(shè)計(jì)、電氣設(shè)計(jì)、給排水設(shè)計(jì)......并且必須留下詳細(xì)的文檔以便使用過程中的維護(hù)。8工程方案書施工質(zhì)量方案書建設(shè)標(biāo)準(zhǔn)和標(biāo)準(zhǔn)施工手冊……復(fù)雜的建筑

9簡單的程序#include<stdio.h>

intmain()

{

printf(“helloworld!〞);//輸出helloworld！

return0;

}簡單的程序，不需要設(shè)計(jì)，直接動手，就可完成。〔嚴(yán)格地說，設(shè)計(jì)過程必不可少，只是迅速地在頭腦中完成了〕10復(fù)雜的程序?yàn)閷?shí)現(xiàn)復(fù)雜的功能需求，應(yīng)用程序具有相當(dāng)復(fù)雜的結(jié)構(gòu)；功能模塊眾多；數(shù)據(jù)結(jié)構(gòu)繁瑣；代碼量巨大。這樣的程序由工程團(tuán)隊(duì)合作完成，無法依靠個人的簡單構(gòu)思完成設(shè)計(jì)并編碼；同時，程序的后期維護(hù)也需要詳細(xì)文檔的支撐。11圖書館信息管理系統(tǒng)--需求分析程序建造的目的是滿足用戶的需求，所以程序設(shè)計(jì)的依據(jù)就是用戶的需求。為準(zhǔn)確把握用戶的需求，需要在與用戶的充分溝通根底上，建立需求文檔；并通過有用戶參與的評審?fù)ㄟ^后，成為設(shè)計(jì)的依據(jù)。12圖書館信息管理系統(tǒng)--功能設(shè)計(jì)設(shè)計(jì)的依據(jù)是經(jīng)過了評審的需求說明書。在實(shí)際建造程序前，應(yīng)該首先完成設(shè)計(jì)工作。設(shè)計(jì)工作是以一種可評審的方式完成系統(tǒng)的邏輯建造工作。設(shè)計(jì)工作從抽象層次較高的系統(tǒng)架構(gòu)設(shè)計(jì)開始，包括功能的設(shè)計(jì)、數(shù)據(jù)結(jié)構(gòu)的設(shè)計(jì)、部署的設(shè)計(jì)、界面的設(shè)計(jì)等。13圖書館信息管理系統(tǒng)--物理模型設(shè)計(jì)14圖書館信息管理系統(tǒng)—局部邏輯模型設(shè)計(jì)15圖書館信息管理系統(tǒng)--數(shù)據(jù)設(shè)計(jì)16圖書館信息管理系統(tǒng)--過程設(shè)計(jì)17圖書館信息管理系統(tǒng)--界面設(shè)計(jì)18軟件開發(fā)是工程化的過程軟件生產(chǎn)同其他行業(yè)，例如建筑行業(yè)一樣需要工程化、規(guī)模化、自動化、標(biāo)準(zhǔn)化。軟件生產(chǎn)過程中也有中間環(huán)節(jié)〔設(shè)計(jì)物理模型、邏輯模型、界面、數(shù)據(jù)、過程等等〕，也可以隨時進(jìn)行質(zhì)量檢查。軟件生產(chǎn)中也需要有高效的工具。軟件生產(chǎn)人員也可以細(xì)分為：分析員、設(shè)計(jì)員、程序員、測試員、操作員、維護(hù)人員等等。軟件生產(chǎn)低水平的手工作坊不能適應(yīng)當(dāng)今的大規(guī)模軟件開發(fā)需求……怎么辦？軟件工程化開發(fā)與維護(hù)！19軟件工程課程的學(xué)習(xí)目標(biāo)掌握軟件工程的根底知識和理論，對軟件工程學(xué)有一個全貌的了解；熟悉軟件工程開發(fā)和維護(hù)的一般過程；掌握軟件需求分析、設(shè)計(jì)、編碼和測試等階段的主要思想和技術(shù)方法；通過學(xué)習(xí)及課程設(shè)計(jì)，真正運(yùn)用和體會軟件工程的思想方法，轉(zhuǎn)變對軟件開發(fā)的認(rèn)識：從個人的單純編程活動轉(zhuǎn)移到進(jìn)行系統(tǒng)分析與設(shè)計(jì)方面上來轉(zhuǎn)變思維定式：程序員―系統(tǒng)工程師〔系統(tǒng)分析員〕20教材軟件工程課程考評課程成績=平時成績X60%+期末成績X40%平時成績=出勤X30%+作業(yè)X40%

+期中考試X30%期末成績=期末閉卷考試成績X100%22Chapter1

IntroductiontoSoftwareEngineeringContent

What'ssoftwareWhat'sagoodsoftwareSoftwarecharacteristicsSoftwarecategoriesSoftwareevolutionSoftwarecrisisWhat'sSESoftwareMyths23SoftwareDefinition24WhatisSoftware?Softwareisasetofitemsorobjectsthatforma“configuration〞thatincludesProgramsDocumentData…SoftwareDefinitionWhatisSoftware?Computerprograms(Instructions)

–whenexecutedprovidedesiredfeatures,function,andperformance;Datastructures–enabletheprogramstoadequatelymanipulateinformation;Documentation(Descriptiveinformation)

–describetheoperationanduseoftheprograms.25SoftwareDocumentationImprovethevisibilityonsoftwareprocessDevelopmoreefficientlyMaketheoutcomeatendofonecertainperiodfordevelopersRecordtechnicalinformationduringthedevelopmentProvideinformationonsoftwareoperation,maintenance,andtrainingHelpuserstounderstandsoftwarefunctions26SoftwareDocumentationFeasibilitystudyreport 可行性研究報(bào)告Softwarerequirementspecification 軟件需求標(biāo)準(zhǔn)Datarequirementspecification 數(shù)據(jù)需求標(biāo)準(zhǔn)Projectdevelopmentplan 工程開發(fā)方案Testingplan 測試方案Conceptual(Architectural)designdocument 架構(gòu)設(shè)計(jì)文件Detaileddesigndocument 詳細(xì)設(shè)計(jì)文檔Databasedesigndocument 數(shù)據(jù)庫設(shè)計(jì)文檔Moduledevelopmentfiles 模塊開發(fā)文件Usermanual 用戶手冊Softwareinstructionmanual 軟件說明書Testinganalysisreport 測試分析報(bào)告Developmentmonthlyreport 開展月度報(bào)告Projectdevelopmentsummaryreport 工程開發(fā)總結(jié)報(bào)告27WhatIsGoodSoftware?PerspectiveonQualityThetranscendentalview:qualityissomethingwecanrecognizebutnotdefineTheuserview:qualityisfitnessforpurposeThemanufacturingview:qualityisconformancetospecificationTheproductview:qualitytiedtoinherentproductcharacteristicsThevalue-basedview:dependsontheamountthecustomersiswillingtopayforit28WhatIsGoodSoftware?GoodsoftwareengineeringmustalwaysincludeastrategyforproducingqualitysoftwareThreewaysofconsideringqualityThequalityoftheproductThequalityoftheprocessThequalityoftheproductinthecontextofthebusinessenvironment29WhatIsGoodSoftware?QualityoftheProductUsersjudgeexternalcharacteristics(e.g.,correctfunctionality,numberoffailures,typeoffailures)Designersandmaintainersjudgeinternalcharacteristics(e.g.,typesoffaults)ThusdifferentstakeholdersmayhavedifferentcriteriaNeedqualitymodelstorelatetheuser’sexternalviewtodeveloper’sinternalview30WhatIsGoodSoftware?QualityModels31JimMcCall〔1977〕BarryW.Boehm〔1978〕FURPS/FURPS+R.GeoffDromeyISO/IEC9126〔1993〕ISO/IEC25010〔2023〕McCall32WhatIsGoodSoftware?QualityoftheProcessQualityofthedevelopmentandmaintenanceprocessisasimportantastheproductqualityThedevelopmentprocessneedstobemodeledModelingwilladdressquestionssuchasWheretofindaparticularkindoffaultHowtofindfaultsearlyHowtobuildinfaulttoleranceWhatarealternativeactivities33WhatIsGoodSoftware?QualityoftheProcessModelsforprocessimprovementSEI’sCapabilityMaturityModel(CMM)ISO9000SoftwareProcessImprovementandCapabilitydetermination(SPICE)34WhatIsGoodSoftware?QualityintheContextoftheBusinessEnvironmentBusinessvalueisasimportantastechnicalvalueBusinessvalue(inrelationshiptotechnicalvalue)mustbequantifiedAcommonapproach:returnoninvestment(ROI)ROIisinterpretedindifferentterms:reducingcosts(effortsandresources),predictingsavings,improvingproductivity.35Software’sDualRolesSoftwareisproduct

Deliverscomputingpotential

Produces,manages,acquires,modifies,displays,or

transmitsinformationSoftwareisavehiclefordeliveringaproduct

Supportsordirectlyprovidessystemfunctionality

Controlsotherprograms(e.g.,anoperatingsystem)

Effectscommunications(e.g.,networkingsoftware)

Helpsbuildothersoftware(e.g.,softwaretools&environment)SoftwareprovidesinformationSoftwaredevelopersfacemanyproblems36SoftwareCharacteristics37Softwareisengineeredordeveloped,notmanufacturedinthetraditionalsense.idealizedcurveFailurerateTimeSoftwaredoesnotwearoutinthesamesenseashardwareSoftwareCharacteristics38SoftwareisdeterioratingMostseriouscorporationscontrolandconstrainchanges.Mostsoftwareiscustombuilt,andcustomerneverreallyknowswhatshe/hewants

SoftwareCategories39SystemSoftware:ServiceotherprogramsCompilers,editorsandfilemanagementutilitiesOperatingsystemcomponents,drivers,networkingsoftwareApplicationSoftware:standaloneprogramsthatsolveaspecificbusinessneedPoint-of-saletransactionprocessingFlightbookingsystemEngineering/ScientificSoftwareNumericalcomputinginphysics,chemistry,biology.Computer-aideddesign,systemsimulationandotherinteractiveapplicationsEmbeddedSoftwarekeypadcontrolforhomeappliancesdigitalfunctionsinanautomobile,e.g.fuelcontrol,brakingsystemsProduct-lineSoftware:DesignedtoprovideaspecificcapabilityforuseWordprocessing,spreadsheets,computergraphicsMultimedia,entertainment,databasemanagementWeb-applicationsE-commerce,B2B,B2P,Blog,TwitterAISoftwareexpertsystem,artificialneuralnetworks,theoremproving,gameplaying,etc.SoftwareNewCategories40Ubiquitouscomputing:wirelessnetworking3GorWi-FiorientedcomputingNetsourcingWebasacomputingengineOpensource“Free〞sourcecodeopentothecomputingcommunityFurthermore…SoftwareNewCategories41DataMiningCloudComputingCognitiveMachines

Softwarefornanotechnologies

LegacySoftware〔使用時間過長而難以維護(hù)的軟件〕

Whymustitchange?Softwaremustbeadaptedtomeettheneedsofnewcomputingenvironmentsortechnology.Softwaremustbeenhancedtoimplementnewbusinessrequirements.Softwaremustbeextendedtomakeitinteroperablewithothermoremodernsystemsordatabases.Softwaremustbere-architectedtomakeitviablewithinanetworkenvironment.

42SoftwareEvolution4319601970198019902000ProgrammingProgramSystemSWOOSESoftwareCrisisTheterm"softwarecrisis"hasbeenusedsincethelate1960s,describethoserecurringsystemdevelopmentproblems:Projectsrunningover-budget.Projectsrunningover-time.Softwarewasveryinefficient.Softwarewasoflowquality.Softwareoftendidnotmeetrequirements.Projectswereunmanageableandcodedifficulttomaintain.(difficulttomodify,addnewfunction,reuse)Softwarewasneverdelivered.44SoftwareCrisis45SoftwareCrisisCasesReportedCaliforniadepartmentofmotorvehiclesspentover$43milliononasystemtomergethestate’sdriverandvehicleregistrationsystems.Systemwasabandonedwithouteverbeingused.Afailed$165millioneffortbyAmericanAirlinestolinkitsflight-reservationsoftwarewiththereservationsystemsofMarriott,Hilton,etc.InMarch,1989,ArthurAndersenreportedMorethan$300billionperyearisspentoncommercialsoftwareactivitiesintheUnitedStatesOnly8%resultsinsoftwarethatisdeliveredANDworks46SoftwareCrisisWhatarethereasonsforthesoftwarecrisis?RollingbaselineShorterdevelopmentcyclesarebusinessrequirementsInitialrequirementsusuallypoorlydefined47SoftwareCrisisWhatarethereasonsforthesoftwarecrisis?SoftwarecomplexityBusinesssoftwaredemandsareincreasingNooneunderstandstheentiresystemLegacysystemsmustbemaintained,buttheoriginaldevelopersaregone48SoftwareCrisisWhatarethereasonsforthesoftwarecrisis?3.FailuretomanageriskThewaterfalllifecyclecandelayproblemidentificationThereisnoproofthatthesystemwillrununtillateinthelifecycleTheresultismaximumrisk49SoftwareCrisisSoftwaredevelopmentisanengineeringproject:Goodorganization

RigorousmanagementMultipledeveloperscollaboration

Efficienttechnique,

methods

andtools

50SoftwareCrisisSoftwaredevelopmentisanengineeringproject:Solutionisusingtechniques(includingmethodsandtools),togetherwithorganizationandmanagement——SoftwareEngineering

SoftwareEngineering:anemergingdisciplinefocusonstudyhowtowelldevelopandmaintainsoftwarefromtechniqueandmanagementperspectives.51BadExperiencesinSECustomersaretoobusytospendtimeworkingwithanalystsordevelopersontheirrequirements.Usersurrogatessuchasmarketingormanagersclaimtospeakfortheuser,buttheyreallydon’t.Customersclaimalltherequirementsarecritical,sotheydonotprioritizethem.Developersencounterambiguitiesandmissinginformationwhencoding,sotheyhavetoguess.52BadExperiencesinSECustomerssignoffontherequirementsandmakechangescontinuously.Theprojectscopeincreasesasrequirementschanges,butscheduleslipsbecausenomoreresourcesareprovided.Functionality/Featuresarerequested,butneverused.Thespecificationissatisfied,butthecustomerisnot.53BadExperiencesinSESystemdesignisnotwelldocumentedandtheoriginaldesignerhaslefttheteam;Sourcecodeisnotwellformatted,whichintroducesextramaintenancecost;Latefoundbugsresultinbudgetoverrun;Customerscomplainsofleakedbugs;Systembecomesmoreandmorecomplexwhileusersatisfactionisgettingdown.Scheduleslipping,budgetburntout.54So,whereisthewayout?Needsystematicalanddisciplinedwaystosolvesuchproblems,tomaketheprojectasuccess!That’swhypeoplestudymethodologies,createtoolstomakeSEaprofession.55What‘sSE？Thetermsoftwareengineeringfirstappearedinthe1968NATOSoftwareEngineeringConference,andwasmeanttoprovokethoughtregardingtheperceived“software

crisis"atthetime.SEisan

engineeringdiscipline

thatisconcernedwithallaspectsofsoftwareproductionfromtheearlystagesofsystemspecificationthroughtomaintaining

thesystemafterithasgoneintouse.ConcernedwithALLaspectsofsoftwareproductionNotjusttechnicalprocess

ofdevelopment.Also

projectmanagement

andthe

developmentoftools,methods

etc.tosupportsoftwareproduction.56Definitionof“SE〞inIEEETheapplicationofasystematic,disciplined,quantifiableapproachtothedevelopment,operationandmaintenanceofsoftware;thatis,theapplicationofengineeringtosoftware.Thestudyofapproachesasin1.EssenceofSE:Max.GainwithMin.Cost!57What‘sSE？SoftwaredevelopmentprocessFrameworkofaseriesoftasksforgettinghighqualifiedsoftware,whichdefinestheworkflowtofinishthesetasksSoftwareengineeringmethodsTechnicalmethodstofinishalldevelopingtasksSoftwaredevelopmenttoolsProvidemethodswithautomaticorsemi-automaticsoftwaresupportingenvironment58IdealGoalsofSoftwareEngineering

Correctness:NoerrorsUsefulness:DoeswhatwewantittodoMinimalproductioneffortAndtherefore:minimalcost,minimaltime,maximumprofitMinimalmaintenanceeffort59FrequentlyAskedQuestionsaboutSEA:Whatissoftware?Q:Computerprograms,dataandassociateddocumentation.Softwareproductsmaybedevelopedforaparticularcustomerormaybedevelopedforageneralmarket.A:Whataretheattributesofgoodsoftware?Q:Goodsoftwareshoulddelivertherequiredfunctionalityandperformancetotheuserandshouldbemaintainable,dependable,efficient,andacceptable.60FrequentlyAskedQuestionsaboutSEA:Whatissoftwareengineering?Q:Softwareengineeringisan

engineeringdiscipline

thatisconcernedwith

allaspectsofsoftwareproduction.A:Whatisthedifferencebetweensoftwareengineeringandcomputerscience?Q:Computersciencefocusesontheoryandfundamentals;softwareengineeringisconcernedwiththepracticalitiesofdevelopinganddeliveringusefulsoftware.61FrequentlyAskedQuestionsaboutSEA:Whatisthedifferencebetweensoftwareengineeringandsystemengineering?Q:Systemengineeringisconcernedwithallaspectsofcomputer-basedsystemsdevelopmentincludinghardware,softwareandprocessengineering.Softwareengineeringispartofthismoregeneralprocess.A:Whatarethekeychallengesfacingsoftwareengineering?Q:Copingwithincreasingdiversity,demandsforreduceddeliverytimesanddevelopingtrustworthysoftware.62FrequentlyAskedQuestionsaboutSEA:Whatarethecostsofsoftwareengineering?Q:Roughly60%ofsoftwarecostsaredevelopmentcosts,40%aretestingcosts.Forcustomsoftware,evolutioncostsoftenexceeddevelopmentcosts.63FrequentlyAskedQuestionsaboutSEA:Whatarethebestsoftwareengineeringtechniquesandmethods?Q:Whileallsoftwareprojectshavetobeprofessionallymanagedanddeveloped,differenttechniquesareappropriatefordifferenttypesofsystem.Forexample,gamesshouldalwaysbedevelopedusingaseriesofprototypeswhereassafetycriticalcontrolsystemsrequireacompleteandanalyzablespecificationtobedeveloped.Youcan’t,therefore,saythatonemethodisbetterthananother.64SoftwareMythsManagementMythsMyth:Ifwegetbehindschedule,wecanaddmoreprogrammersandcatchup.Reality:Softwaredevelopmentisnotamechanisticprocesslikemanufacturing.InthewordsofBrooks“Addingpeopletoalate,softwareprojectmakesitlater〞Peoplecanbeaddedonlyinaplannedandwell-coordinatedmanner.65SoftwareMythsCustomerMythsMyth:Projectrequirementscontinuallychange,butchangecanbeeasilyaccommodatedbecausesoftwareisflexible.Reality:Theimpactofchangevarieswiththetimeatwhichitisintroduced.Astimepasses,costimpactgrowsrapidly.Changecancauseupheavalthatrequiresadditionalresourcesandmajordesignmodification.

66SoftwareMythsPractitioner’sMythsMyth:UntilIgettheprogramrunning,Ihavenowayofassessingitsquality.Reality:Theformaltechnicalreviewisoneofmosteffectivesoftwarequalityassurancemechanisms.Softwarereviewsarea“qualityfilter〞thathavebeenfoundtobemoreeffectivethantestingforfindingcertainclassesofsoftwareerrors.6768ExerciseWhichquestionnolongerconcernsthemodernsoftwareengineeringWhydoescomputerhardwarecostsomuch?Whydoessoftwaretakealongtimetofinish?Whydoesitcostsomu