
SOFTWARE QUALITY: BARBARAKITCHENHAM, National Computing Centre SHARI LAWRENCE PFLEEGER, Systerns/Software, Inc. n the recent past, when bank ; context is important. Errors toler- statements contained errors ated in word-processing software or the telephone network may not be acceptable in control broke down, the general pub- software for a nuclear-power plant. Ilic usually blamed “the computer,” Thus, we must reexamine the i making no distinction between i meanings of “safety-critical” and hardware and software. However, “mission-critic.al” in the context of high-profile disasters and the ensu- software’s contribution to the ing debates in the press are alerting i larger functionality and quality of more people to the crucial nature ! products and businesses. At the of software quality in their every- same time, we must ask ourselves day lives. Before long, we can ex- who is responsible for setting qual- pect increasing public concern ity goals and making sure they are about the pervasiveness of soft- achieved. ware, not only in public services but also in consumer products like i WHAT DOES QUALITY automobiles, washing machines, REALLY MEAN? telephones, and electric shavers. i Consequently, we software profes- ! Most of us are affected by the sionals need to worry about the ! quality of the software we create quality of all our products - from because our organization’sviability large, complex, stand-alone sys- depends on it. And most software- tems to small embedded ones. ! related tools and methods - in- So how do we assess “adequate” ; cluding those described in IEEE quality in a software product? The Sofcware - claim to assess or im- 12 0740 7459/96/$05 00 0 1996 IEEE JANUARY 1996 Authorized licensed use limited to: The University of Arizona. Downloaded on January 9, 2009 at 13:16 from IEEE Xplore. Restrictions apply. prove software quality in some way. So for overall product quality. + The manujicturing view sees qual- we must question what we and our cus- In their more general questioning of ity as conformance to specification. tomers mean by software quality. quality goals and techniques, Roger + Theproduct view sees quality as tied A good definition must let us mea- Howe, Dee Gaeddert, and Maynard to inherent characteristics of the product. sure quality in a meaningful way. Howe pointed out that most quality ini- + The value-based view sees quality as Measurements let us know if our tech- tiatives either fail (by drowning in a sea dependent on the amount a customer is niques really improve our software, as of rhetoric) or cannot demonstrate suc- willing to pay for it. well as how process quality affects prod- cess because no financial retun can be uct quality. We also need to know how identified.’ In this special issue, we ques- Transcendental view. This view of soft- the quality we build in can affect the tion software quality in the same way. ware quality is much like Plato’s descrip- product’s use after delivery and if the in- We consider the meaning of software tion of the ideal or Aristotle’s concept of vestment of time and resources to assure quality, how we assess it, and whether form. Just as every table is different but high quality reap higher profits or larger the steps we are taking to improve it are each is an approximation of an ideal market share. In other words, we want to really worthwhile. table, we can think of software quality as know if good software is good business. something toward which we strive as an Recent articles have raised this ques- VIEWS OF SOFTWARE QUALITY ideal, but may never implement com- tion, but the answer is still far from clear. pletely. When software gurus exhort us Still, most people believe that quality is In an influential paper examining to produce products that delight users, important and that it can be improved. views of quality, David Gamin studied this delight represents the strived-for Companies and countries continue to how quality is perceived in various do- “recognition” in the transcendental def- invest a great deal of time, money, and mains, including philosophy, economics, inition of quality. effort in improving software quality. But marketing, and operations management.’ we should try to determine if these na- He concluded that “quality is a complex User view. Whereas the transcenden- tional initiatives have directly affected and multifaceted concept” that can be de- tal view is ethereal, the user view is more and improved software quality. The an- scribed from five different perspectives. concrete, grounded in product charac- swer may depend on how you approach + The transcendental view sees quality teristics that meet the user’s needs. This quality improvement. Some companies as something that can be recognized but view of quality evaluates the product in take a product-based approach, while not defined. a task context and can thus be a highly others focus on process; both strategies + The user view sees quality as fitness personalized view. In reliability and per- have led to Malcolm Baldridge awards for purpose. formance modeling, the user view is in- EOFTWARE QUALITY SURVEY IASLyear we invited readers lo air dtictirity a11d ninc thought protluctivit\.. their view on soft~rare quality by coni- n:1s of equal or greater importance. plctiiig a short questiomiairc. As III,II~II~~I~~IIII~~I~~i iw(17 and 13, rcspcc- Quality issues. \.\:e asked rcsponilcnts proniisetl?we report the results of the ti\L*l! I. I loi\c~\.ci.,14 respondents sus- to select thrce cplity issues from a list sui-vey here. \,\’e thank those who coiii- qcstcd definitions that covered two or and rank them in terms of inipc)rtance. pleted the questionnaire. Of the 27 re- more different views. Other vicwpoiii ts Mi. used :i simple ranking ordcr for thc spondents, 1; uwe froni rhe US, live included product (nine reslioiidents), igning three points to the froni Europe, and from Asia - sim- tr,i n scende n tal (thrcc), and \,a I ue (fire). fiw item niarkcd as most inip-fiint, two Of the 27 respondents, I8 strongly ilar to the distribution of lb.’/*,‘f;.’ii!~mm points to those marked ;is next most im- agreed and seven aprccd that software reiiders. Respondents l~ackgroundcxpc- portant, and one point to thc third inwr rience was mixed and soiuc indivitluals qtdit~constituted a prol)leiii. \Vhcn important issuc. marked several categor asked if a cl~~iIit!-in:iiiageiiieiitsystciii Respondents ranked speci$ing qu;iI- iii could qidity nine marks in developmeni and seven alone solve tlic problein. ity requirernents ohjcctively as most ini- rescarch. Of‘the 12 who markecl “other,” 18 disagreed, thrcc :igrcetl, nnd t~vore- portant (28 points), followed 11. scttirig five \\rote in “qmalin-asstirmce.” S~l~Jlld~ll~SSLl.Ollgl~ >lgrCcd. 111) a cl~i~ilit\:-nianagcriieiitspsteni (20 :\Ithough this is ncithcr ;I large nor Opinions were inixc ’ .. I., ....‘a. :‘.: . points). This implics that respoideiits qualit!! is nioic or Icss ib..i .:I,, : : :I rcprcsenrati\ e saniple, we hope that thc rate a UhIS as necessary - but not suf- !;OLI time to market: 17 thought time to responses will rnaltc think ahout ficient - for addressing qiialiri,. inarket oFeqti:il or greater inipor- your own perspective on quality and cx- was :Ichieving operational quality that arniuc how qii;ility is effected hy the tance than quality iiii(l wx’ii ilioiigli~ inccts requirements U as ~.:inl;etlthird tiest-liracticeactivities !~LIirnplenient. qii;lli t!, n as no re ii I I I)( i11;i 111 !I 11 I’CY Iiii(I (18 points), follon-ed b!; mcasiiring no opinion). On the relative vdue of Views of quality. ;isLcd qunlity achievcincnts (1 7 points), and \.Vc respon- quality ancl protlucti\in; 17 thought tlcnts to suggest thcir own quality cleti- agreeing with thc ctistoiiier on nhat qixility \vas inore important tlian pro- nitions :ind then assessed thein against qualit!, iiieans (1 5 pints). IEEE SOFTWARE Authorized licensed use limited to: The University of Arizona. Downloaded on January 9, 2009 at 13:16 from IEEE Xplore. Restrictions apply. GUEST EDITORS’ INTRODUCTION herent, since both methods assess prod- nal quality affect the product’s use. Some influences how we define it. But we also uct behavior with respect to operational researchers have developed models to want to be able to measure quality so we profiles (that is, to expected functionality link the product view to the user view. can establish baselines, predict likely and usage patterns). Product usability is quality, and monitor improvement. also related to the user view: in usability Value-based view. Different views can be Here, too, perspective influences our laboratories, researchers observe how held by different groups involved in soft- choice. Users assess software-product users interact with software products. ware development. Customersor market- quality in terms of their interaction with ing groups typically have a user view, re- the final product. Product attributes that Manufacturing view. The Manufactur- searchers a product view, and the contribute to user satisfaction are a mix- ing view focuses on product quality production department a manufacturing ture of during production and after delivery. view. If the difference in viewpoints is not + the product’s functions, which are This view examines whether or not the made explicit, misunderstandings about either present or absent; product was constructed “right the first quality created during project initiation + the product’s nonfunctional quali- time,” in an effort to avoid the costs as- are likely to resurface as (potentially) major ties (its behavior), which is measurable sociated with rework during develop- problems during product acceptance. within some range; and ment and after delivery. This process These disparate views can comple- + the constraints that determine if a focus can lead to quality assessment ment each other in early phases.
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages8 Page
-
File Size-