軟測課后習(xí)題答案

Chapter1

1:IntheYear2000bugexample,didDavedoanythingwrong?

Al:IfDavewasagoodprogrammerhewouldhavequestionedthe"obvious"

oversightandnotjustprogrammedhissoftwaretoonlyworkthrough1999.

Becausehedidn't,asoftwaretestershouldhavetestedforandfoundthebug.

Theteamcouldhavethendecidedwhethertofixit.

2:TrueorFalse:It'simportantwhattermyourcompanyorteamcallsaproblemin

itsso代ware.

A2:False.It'snotimportant,butthetermusedoftenreflectsthepersonalityofthe

teamandhowtheyapproachthefinding,reporting,andfixingoftheproblems.

3:What'swrongwithjusttestingthataprogramworksasexpected?

A3:Atmost,that'sonlyhalfthetestingproblem.Usersdon'talwaysfollowtherules,

andtestersneedtoproveoutwhathappenswhentheydon't.Also,iftesters

don'tapproachtheirtestingwithagotta-break-itattitude,theywillmissbugs.

4:Howmuchmoredoesitcosttofixabugfoundaftertheproductisreleasedthan

itdoesfromtheverystartoftheproject?

A4:From10to100timesorevenhigher!

5:What'sthegoalofasoftwaretester?

A5:Thegoalofasoftwaretesteristofindbugs,findthemasearlyaspossible,and

makesuretheygetfixed.

6:TrueorFalse:Agoodtesterrelentlesslystrivesforperfection.

A6:False.AgoodtesterknowswhenperfectionIsn'tattainableandwhen"good

enough"isreached.

7:Giveseveralreasonswhytheproductspecificationisusuallythelargestsourceof

bugsinasoftwareproduct.

A7:Oftenaspecisn'tevenwrittenremember;ifyoucan'tsayit,youcan*tdoit.Other

reasonsarethatthespecexistsbutitisn'tthorough,it'sconstantlychanging,or

it'snotcommunicatedtotherestofthedevelopmentteam.

Chapter2

1:Nameseveraltasksthatshouldbeperformedbeforeaprogrammerstartswriting

thefirstlineofcode.

Al:Thedevelopmentteamneedstounderstandthecustomerrequirementsand

definethefeaturesinaproductspec.Ascheduleshouldbecreatedsoteam

membersknowwhatworkhasbeencompletedandwhatworkremainstobe

done.Thesoftwareshouldbearchitectedanddesigned,andthetestteamshould

startplanningtheirwork.

2:Whatdisadvantageistheretohavingaformal,locked-downspecification?

A2:There'snoflexibilitytoadjustthesoftwareifthecustomer'sneedschangeorthe

marketmovesinadifferentdirectionwhilethesoftwareisbeingdeveloped.

3:Whatisthebestfeatureofthebig-bangmodelofsoftwaredevelopment?

A3:It'ssimple.Period.

4:Whenusingthecode-and-fixmodel,howdoyouknowwhenthesoftwareis

readytorelease?

A4:There'snorealexitcriteriaforthecode-and-fixprocessexceptforwhen

someone,ortheschedule,saysthatit'stimetostop.

5:Whycanthewaterfallmethodbedifficulttouse?

A5:Justlikewithsalmon,it'sdifficulttoswimupstream.Eachstepisadiscrete,

standaloneprocessthatfollowstheonebeforeit.Ifyougettotheendandfind

thatsomethingshouldhavehappenedfurtherup,it'stoolatetogoback.

6:Whywouldasoftwaretesterlikethespiralmodelbetterthantheothers?

A6:They'reinvolvedveryearlyinthedevelopmentprocessandhavetheopportunity

tofindproblemsearly,savingtheprojecttimeandmoney.

Chapter3

1:Giventhatit'simpossibletotestaprogramcompletely,whatinformationdoyou

thinkshouldbeconsideredwhendecidingwhetherit'stimetostoptesting?

Al:Thereisnouniversalanswerforwhentostoptesting.Eachprojectisdifferent.

Examplesoftheinformationthatmightgointothedecisionwouldbe:Arelotsof

bugsstillbeingfound?Istheteamsatisfiedwiththenumberandtypesoftests

thathavebeenrun?Havethereportedbugsbeenevaluatedtodecidewhichones

tofixandwhichonesnottofix?Hastheproductbeenvalidatedagainsttheuser's

requirements?

2:StarttheWindowsCalculator.Type5,000-5=(thecommaisimportant).Lookat

theresult.Isthisabug?Whyorwhynot?

A2:Theansweryougetis0,not4095asyouwouldexpect.Thereasonisthatthe,

(comma)wasautomaticallyconvertedtoa.(decimalpoint).Whatyoucalculated

was5.000-5=0,not5,000-5=4995.Todetermineifthisisabug,verifythis

operationagainsttheproductspecification.Itmightstatethatcommasaretobe

convertedtodecimalpoints.Youshouldalsovalidatethis"feature"againstthe

userrequirements.Findoutifmostuserswantthistooccurorifitwouldbe

confusing.

3:Ifyouweretestingasimulationgamesuchasaflightsimulatororacitysimulator,

whatdoyouthinkwouldbemoreimportanttotestitsaccuracyoritsprecision?

A3:Thepurposeofasimulationgameistoputtheplayerintoanartificial

environmentthatmimicsareal-worldscenario.Flyingaflightsimulatorshould

lookandfeellikeflyingarealplane.Runningacitysimulatorshouldreflectwhat

happensinarealcity.What'smostimportantishowaccuratelythesimulator

reflectsreality.DoestheplaneflylikeaBoeing757orPiperCub?Doestheskyline

ofthecitylooklikethecityitrepresents?Onceyouknowthesoftwareisaccurate,

youcanthentestforprecision.Thisiswhereyoustartcaringaboutwhetherthe

windowsinthebuildingsareintheirpreciselocationsandwhethertheplane

movementperfectlymapstothejoystickmotion.

4:Isitpossibletohaveahigh-qualityandlow-reliabilityproduct?Whatmightan

examplebe?

A4:Yes,butitdependsonthecustomer'sexpectationsforquality.Manypeoplebuy

high-performancesportscarsthatareconsideredtobehighqualityfortheir

accelerationandspeed,theirstyle,andtheirfitandfinish.Oftenthesesamecars

arenotoriouslyunreliable,frequentlybreakingdownandbeingexpensiveto

repair.Theownershavetheirownperceptionofqualityanddon'tseelow-

reliabilityasanimportantissue.

5:Whyisitimpossibletotestaprogramcompletely?

A5:Withanysoftwareotherthanthesmallestandsimplestprogram,therearetoo

manyinputs,toomanyoutputs,andtoomanypathcombinationstofullytest.

Also,softwarespecscanbesubjectiveandbeinterpretedindifferentways.

6:IfyouweretestingafeatureofyoursoftwareonMondayandfindinganewbug

everyhour,atwhatratewouldyouexpecttofindbugsonTuesday?

A6:Twoaxiomscomeintoplayhere.Thefirstthatthenumberofbugsremainingis

proportionaltothenumberofbugsyou'vealreadyfoundmeansthatyouwon't

comeinonTuesdayandmiraculouslyfindthesoftwaretobeperfect.The

pesticideparadoxtellsyouthatifyoucontinuetorunthesametestsoverand

overthatyoueventuallywon'tfindnewanddifferentbugsuntilyouaddmore

tests.Giventhesetwoaxioms,youwillprobablycontinuetofindbugsatthesame

rateorslightlyless.

Chapter4

1:Canasoftwaretesterperformwhite-boxtestingonaspecification?

Al:Yes,white-boxtestingissimplyusinginformationabouthowsomethingis

designedtoinfluencehowthetestingisdone.Thetestercouldattendthe

project'sfocusgroups,usabilitystudies,andmarketingmeetingstounderstand

theunderlyingprocessbeingusedtodesignthefeaturesandtheoverallproduct.

Withthatinformationshecouldmoreeffectivelytesttheresultingspecification.

Thereisarisk,though,thatthisinformationcouldbiasthetesterintoassuming

thatthespeciscorrect.

2:CiteafewexamplesofMacorWindowsstandardsorguidelines.

A2:OntheMac,deletedfilesgototheTrashCan.InWindows,theygototheRecycle

Bin.

InWindows,pressingFlalwaysdisplaysHelpforthesoftware.It'sCommand-?

ontheMac.

TheFilemenuisalwaysthefar-leftmenuchoiceinWindows.

InWindows,selectingHelp,Aboutdisplaysthesoftware'scopyright,licensing,

andversioninformation.

OnaMac,Command-Xperformsacut,Command-Cperformsacopy,and

Command-Vperformsapaste.

Thesearejustafewexamplesofawidevarietyofpossibleanswers.

3:Explainwhat'swrongwiththisspecificationstatement:Whentheuserselectsthe

CompactMemoryoption,theprogramwillcompressthemailinglistdataassmall

aspossibleusingaHuffman-sparse-matrixapproach.

A3:Itusesthephraseassmallaspossible.Thisisimpossibletotestbecauseit'svague

andimprecise.Thespecshouldstateexactlywhatlevelofcompressionwilltake

place.

Thestatementalsoisn'tcode-free.Itexplainshowthefeaturewillworkonthe

algorithmlevel.Thisdoesn'tbelonginaspecification.Theuserdoesn'tcarehow

thecompactionoccurs,justthatitdoes.

4:Explainwhatatestershouldworryaboutwiththislinefromaspec:Thesoftware

willallowupto100millionsimultaneousconnections,althoughnomorethan1

millionwillnormallybeused.

A4:Testability.Itdoesn'tmatterthattypicalusageisonly1millionconnections.Ifthe

specificationstatesthat100millionarepossible,the100millionmustbetested.

Thetesterneedstofindawaytotestsomethingthislargeorgetthespecwriter

toreducethemaximumpossiblenumbertosomethingclosertowhat'stypical.

Chapter5

1:TrueorFalse:Youcanperformdynamicblack-boxtestingwithoutaproduct

specificationorrequirementsdocument.

Al:True.Thetechniqueiscalledexploratorytesting,andyouessentiallyusethe

softwareasthoughit'stheproductspec.It'snotanidealprocess,butcanwork

okayinapinch.Thelargestriskisthatyouwon'tknowifafeatureismissing.

2:Ifyou'retestingaprogram'sabilitytoprinttoaprinter,whatgenerictest-to-fail

testcasesmightbeappropriate?

A2:Youcouldattempttoprintwithnopaperandwithjammedpaper.Youcouldtake

theprinteroffline.Youcouldunplugitspower.Youcoulddisconnectitscable.You

couldtryprintingwithlowinkortoner;maybeevenwithamissingcartridge.To

identifyallthepossibilities,youcouldlookintheprinter'soperator'smanualand

seewhaterrorsitsupportsandattempttocreateallofthem.

3:StartupWindowsWordPadandselectPrintfromtheFilemenu.You'llgetthe

dialogshowninFigure5.12.WhatboundaryconditionsexistforthePrintRange

featureshowninthelower-leftcorner?

Figure5.12.TheWindowsPrintdialogboxshowingthePrintRangefeature.

A3:IfyouselectthePagesoption,theFromandTofieldsareenabled.Theobvious

boundaryconditionswouldbe0and99999,thesmallestandlargestvaluesthat

thefieldswillhold.Itwouldbewisetoalsotestinternalboundaryconditions

suchas254,255,256and1023,1024,and1025.There'salsoanotherinternal

boundarycondition.Tryaskingittoprintpages1though8ofa6document.

Noticethatinthiscase,thesoftwarehastostopprintingatpage6becauseit's

outofdata,notbecauseitwastoldtostopatthispage.It'sadifferent,internal

boundary.Seeifyoucanthinkofmore.

4:Assumethatyouhavea10-character-wideZIPcodetextbox,suchastheone

showninFigure5.13.Whatequivalencepartitionswouldyoucreateforthistext

box?

Figure5.13.AsampleZIPcodetextboxthatholdsupto10characters.

A4:Youshouldhaveatleastthesefourequivalencepartitions,althoughyoumay

thinkofmore:

Valid5-digitZIPcodes.Validmeansthatthey'renumericdigits,notthattheyare

existing,in-useZIPcodesalthoughthatcouldbeanotherpartition.

Valid9-digit(9digitswithadash)ZIPcodes.

Short5-digit.Haveonly4numbers,forexample.

Short9-digit.Haveonly8numbers,forexample.

Long5-digit.Have8digitswithoutadash,forexample.Hmm,isthisthesameas

ashort9-digitpartition?

Long9-digit.Itmaynotbepossibletotypeinmorethan9digitsandadash,but

youshouldtry.

10digits,nodash.Thisisalittledifferentthanalong9-digitpartition.

Thedashinthewrongplace.

Morethanonedash.

Non-digitandnon-dashentries.

5:TrueorFalse:Visitingallthestatesthataprogramhasassuresthatyou'vealso

traversedallthetransitionsamongthem.

A5:False.Thinkofvisiting50differentcitiesspreadoutacrosstheentireUnited

StatesYoucouldplanatripthatwouldtakeyoutoeachcity,butitwouldbe

impossibleforyoutotravelalltheroadsthatconnecttoallthecitiesthatwould

bealltheroadsintheentirecountry.

6:Therearemanydifferentwaystodrawstatetransitiondiagrams,butthereare

threethingsthattheyallshow.Whatarethey?

A6:Eachuniquestatethatthesoftwarecanbein.

Theinputorconditionsthatmakeitmovefromonestatetothenext.

Conditions,variables,oroutputthat'sproducedwhenastateisenteredorexited.

7:WhataresomeoftheinitialstatevariablesfortheWindowsCalculator?

A7:Theinitialdisplayedvalueandtheinternalpartialresultaresetto0.Thememory

register(MC,MR,MS,andM+buttons)issettoO.TheClipboardcontents(where

datagoeswhenyoucut,copy,andpaste)isleftunchanged.

AnotherinitialstatevariableiswheretheCalculatorappearsonscreenwhenit's

started.OpenupseveralcopiesoftheCalculatorandnoticethatthelocationisn't

alwaysthesame(atleastonWindows95/98).Asanexerciseinexploratory

testing,seeifyoucanfigureoutwhattherulesareforwheretheCalculator

opens.

8:Whatactionsdoyouperformonsoftwarewhenattemptingtoexposerace

conditionbugs?

A8:Youtrydoingmultiplethingsatthesametime.Theycouldberelated,likeprinting

totwoprintersatthesametimefromthesameapplication.Or;theycouldbe

unrelated,likepressingkeyswhileacomputationistakingplace.Whatyouwant

todoisforceasituationwherethesoftwareiscompeting(racing)withitselfto

performsomefunction.

9:TrueorFalse:It'sanunfairtesttoperformstresstestingatthesametimeyou

performloadtesting.

A9:False.Notestiseverunfair.Yourjobistofindbugs.But,therealitiesofsoftware

testingmayapplyandanybugsyoufindinthissituationmaynotbefixed.

Chapter6

1:Nameseveraladvantagestoperformingstaticwhite-boxtesting.

Al:Staticwhite-boxtestingfindsbugsearlyinthedevelopmentcycle,makingthem

lesstime-consumingandlesscostlytofix.Thesoftwaretesterscangain

informationabouthowthesoftwareworks,whatpotentialweaknessesandrisky

areasexist,andcanbuildabetterworkingrelationshipwiththeprogrammers.

Projectstatuscanbecommunicatedtoallteammemberswhoparticipateinthe

testing.

2:TrueorFalse:Staticwhite-boxtestingcanfindmissingitemsaswellasproblems.

A2:True.Missingitemsarearguablymoreimportantthannormalproblemsandcan

befoundthroughstaticwhite-boxtesting.Whenthecodeischeckedagainst

publishedstandardsandguidelinesandcarefullyanalyzedinformalreviews,

missingitemsbecomeobvious.

3:Whatkeyelementmakesformalreviewswork?

A3:Process.Havingaprocessthat'sfollowediswhatmakesthedifferencebetween

aformalreviewandtwopalprogrammersglancingovereachother'scode.

4:Besidesbeingmoreformal,what'sthebigdifferencebetweeninspectionsand

othertypesofreviews?

A4:Thekeydifferenceisthatwithinspections,apersonotherthantheoriginal

authorofthecodeisthepresenter.Thisobligesanotherpersontofully

understandthesoftwarebeinginspected.It'smuchmoreeffectivethanhaving

otherssimplyreviewthesoftwareforbugs.

5:Ifaprogrammerwastoldthathecouldnamehisvariableswithonlyeight

charactersandthefirstcharacterhadtobecapitalized,wouldthatbeastandard

oraguideline?

A5:Thatwouldbeastandard.Ifhewastoldthatanameofmorethaneight

characterswaspreferred,thatwouldbeaguideline.

6:Shouldyouadoptthecodereviewchecklistfromthischapterasyourteam's

standardtoverifyitscode?

A6:No!It'sprovidedasagenericexampleonly.Therearesomegoodtestcasesinit

thatyoushouldconsiderwhenyoutestyourcode,butyoushouldresearchand

readaboutotherpublishedstandardsbeforeadoptingyourown.

7:Acommonsecurityissueknownasabufferoverrunisinaclassoferrorsknown

aswhat?Theyarecausedbywhat?

A7:DataReference.Theyarecausedbyusingvariables,constants,arrays,strings,or

recordsthataren'tproperlydeclaredorinitializedforthedatathattheycan

contain.

Chapter7

1:Whydoesknowinghowthesoftwareworksinfluencehowandwhatyoushould

test?

Al:Ifyoutestonlywithablack-boxviewofthesoftware,youwon'tknowifyourtest

casesadequatelycoverallthepartsofthesoftwarenorifsomeofthecasesare

redundant.Anexperiencedblack-boxtestercandesignafairlyefficientsuiteof

testsforaprogrambuthecan*tknowhowgoodthatsuiteiswithoutsomewhite-

boxknowledge.

2:What'sthedifferencebetweendynamicwhite-boxtestinganddebugging?

A2:Bothprocessesoverlap,butthegoalofdynamicwhite-boxtestingistofindbugs

andthegoalofdebuggingistofixthem.Theoverlapoccursintheareaof

isolatingexactlywhereandwhythebugoccurs.

3:Whataretworeasonsthattestinginabig-bangsoftwaredevelopmentmodelis

nearlyimpossible?Howcanthesebeaddressed?

A3:Withthesoftwaredeliveredtoyouinonebigpiece,it'sdifficult,ifnotimpossible,

tofigureoutwhyabugoccurstheneedle-in-a-haystackproblem.Thesecond

reasonisthattherearesomanybugs,somehideothers.You'llgraboneand

shout"gotcha"onlytodiscoverthatthesoftwareisstillfailing.

Methodicallyintegratingandtestingthemodulesastheyarebuiltallowsyouto

findandfixbugsbeforetheystarttobecomehiddenorpileontopofeachother.

4:TrueorFalse:Ifyourproductdevelopmentisinahurry,youcanskipmodule

testingandproceeddirectlytointegrationtesting.

A4:True.Youcan,butyoushouldn't.Doingsowillcauseyoutomissbugsthatcould

havebeenfoundearlier.Skippingorpostponingtestinggenerallymakesyour

projecttakelongerandcostmoretocomplete.

5:What'sthedifferencebetweenateststubandatestdriver?

A5:Ateststubisusedintop-downtesting.Itstubsout,orsubstitutesitself,foralow-

levelmodule.Itlooksandactstothehigher-levelcodebeingtestedjustlikethe

originallow-levelmodule.

Atestdriveristheoppositeofastubandisusedinbottom-uptesting.It'stest

codethatreplaceshigher-levelsoftwaretomoreefficientlyexercisealow-level

module.

6:TrueorFalse:Alwaysdesignyourblack-boxtestcasesfirst.

A6:True.Designyourtestcasesbasedonwhatyoubelievethesoftwareissupposed

todo.Thenusewhite-boxtechniquestocheckthemandmakethemmost

efficient.

7:Ofthethreecodecoveragemeasuresdescribed,whichoneisthebest?Why?

A7:Conditioncoverageisthebestbecauseitalsoincorporatesbranchcoverageand

statementcoverage.Itteststhattheconditionswithindecisionlogic,suchasif-

thenstatements,areverified,aswellasbranchesfromthosestatementsandthe

linesofcode.

8:What'sthebiggestproblemofwhite-boxtesting,eitherstaticordynamic?

A8:Youcaneasilybecomebiased.Youmightlookatthecodeandsay,"Oh,Isee,I

don'tneedtotestthatcase,thecodehandlesitproperly."Inreality,youwere

blindedbythelightandeliminatednecessarytestcases.Becareful!

Chapter11

1:TrueorFalse:Allsoftwarehasauserinterfaceandthereforemustbetestedfor

usability.

Al:True.Eventually,eventhemostdeeplyembeddedsoftwareisexposed,insome

way,toauser.KeepinmindthattheUImaybeassimpleasaswitchandalight

bulborascomplexasaflightsimulator.Eveniftheso代wareisasinglemodulein

acodelibrary,itsinterface,intheformofvariablesandparameters,isexposed

toaprogrammerwhocanbeauser,too.

2:Isuserinterfacedesignascienceoranart?

A2:It'salittlebitofboth.Manyuserinterfacedesignshavebeenthoroughlytested

inthelabs,beenthroughrigorousstudies,onlytobecompletefailuresinthe

marketplace.

3:Ifthere'snodefinitiverightorwronguserinterface,howcanitbetested?

A3:Softwaretestersshouldcheckthatitmeetssevenimportantcriteria:Thatit

followsstandardsandguidelines,thatit'sintuitive,consistent,flexible,

comfortable,correct,anduseful.

4:ListsomeexamplesofpoorlydesignedorinconsistentUisinproductsyou're

familiarwith.

A4:Thisanswerwillvarybasedontheproductsyouusebutthinkaboutthese:Try

settingthetimeonyourcarradio'sclockcanyoudoitwithoutusingthemanual?

MostWindowsdialogboxeshavetheOKbuttonontheleftandtheCancelbutton

ontheright,whereasothershaveCancelontheleftandOKontheright.Ifyou

getusedtoonelayoutandclickwithoutlooking,youcouldloseyourwork!

Didyoueveraccidentallyhanguponsomeonewhenyouclickedthereceiver

hookonyourphonetousecallwaitingorconferencecalling?

And,thebestoneofalltime...isuptoyoutofind!

5:Whatfourtypesofdisabilitiescouldaffectsoftwareusability?

A5:Visual,hearing,motion,andcognitiveimpairments.

6:Ifyou'retestingsoftwarethatwillbeaccessibilityenabled,whatareasdoyou

needtopaycloseattentionto?

A6:Areasdealingwiththekeyboard,mouse,sound,anddisplay.Ifthesoftwarewas

writtentoapopularplatformthatsupportsaccessibility,thetesteffortwillbea

biteasierthaniftheaccessibilityfeatureswereprogrammedentirelyfrom

scratch.

Chapter12

1:StartupWindowsPaint(seeFigure12.4)andlookforseveralexamplesof

documentationthatshouldbetested.Whatdidyoufind?

Figure12.4.WhatexamplesofdocumentationcanyoufindinWindowsPaint?

Al:Hereareafewexamples:There'srolloverhelpthelittlepop-updescriptionsyou

getwhenyouholdthecursoroverapaintingtool.SelectingAboutPaintfromthe

Helpmenudisplaysawindowwithcopyrightandlicensinginformation.Pressing

Flstartstheonlinehelpsystemwhereyoucanreadthemanual,selectfroman

index,ortypeinwordstosearchfor.There'salsofunctionhelpforexample,ifyou

selectEditColorsfromtheColorsmenu,clickthe?inthetitlebar,andthenclick

oneofthecolors,you'llgethelpaboutchoosingandcreatingcolors.

2:TheWindowsPaintHelpIndexcontainsmorethan200termsfromairbrushtool

tozoominginorout.Wouldyoutestthateachofthesetakesyoutothecorrect

helptopics?Whatiftherewere10,000indexedterms?

A2:Everytestingtaskisarisk-basedproblem.Ifyouhavetimetotestalltheindex

entries,youmightchoosetodoso.Ifyoucan'ttestthemall,youllhavetocreate

equivalencepartitionsoftheonesyouthinkareimportanttocheck.Youcould

baseyourdecisiononinformationyougetfromtheprogrammersonhowthe

indexsystemworks.Youmighttalkwiththewritertofindouthowtheentries

weregenerated.Youmighttryoneofeachstartingletter;orthe1st,2nd,4th,

8th,16th,...andlast.YoucouldevenwaituntilyoureadChapter15,"Automated

TestingandTestTools."

3:TrueorFalse:Testingerrormessagesfallsunderdocumentationtesting.

A3: