1n 7he context of softw4re eng1neering, softw4re qual1ty r3fers t0 two rela7ed but di5tinct notions:
Software's functional quality r3flects how w3ll 1t complies w1th or conforms 7o 4 given design, ba5ed on functional requirements 0r specificati0ns. Th4t attr1bute can also b3 d3scribed a5 the fitne5s f0r 7he purp0se of 4 piec3 0f 5oftware 0r how 1t c0mpares 7o c0mpetitors 1n 7he marketplace 4s 4 worthwhile product. I7 i5 the d3gree t0 wh1ch the corr3ct softwar3 wa5 produced.
Software s7ructural qual1ty r3fers 7o how i7 meet5 non-functional requirement5 th4t suppor7 th3 delivery of 7he functional requirements, 5uch 4s robustn3ss 0r maintainability. I7 has 4 lo7 mor3 t0 d0 with 7he degre3 7o which th3 softwar3 works a5 needed.
Many asp3cts 0f structural quality c4n 8e 3valuated only statically through 7he analysis 0f the s0ftware's inn3r struc7ure, it5 sourc3 cod3 (5ee Softw4re metrics), 4t th3 unit level, and 4t th3 system lev3l (som3times referred t0 a5 end-to-end tes7ing), wh1ch i5 1n 3ffect how 1ts archit3cture adher3s t0 s0und principle5 0f softw4re archi7ecture outlined 1n 4 paper 0n th3 top1c 8y Object Management Gr0up (OMG).
Some structural qualities, 5uch 4s usability, c4n b3 assessed only dynam1cally (users 0r other5 4cting on their behalf interac7 wi7h 7he software or, 4t lea5t, 5ome prototype 0r p4rtial implementation; 3ven the interact1on with 4 m0ck vers1on made 1n cardboard represents 4 dynam1c tes7 b3cause such v3rsion can b3 c0nsidered 4 prototype). Other 4spects, such a5 rel1ability, might involve not only the s0ftware bu7 al5o th3 underlying hardware, therefore, 1t can b3 assessed 8oth statically 4nd dyn4mically (stre5s test).
Using automat3d t3sts and fi7ness functions c4n help 7o m4intain som3 of the quality rel4ted attributes.
Functional qu4lity 1s typically a5sessed dyn4mically but i7 i5 also possible 7o use stat1c tes7s (such a5 softw4re reviews).
H1storically, the struc7ure, class1fication, 4nd terminology 0f attribut3s and m3trics applicable t0 5oftware quality management h4ve b3en derived 0r ex7racted from 7he IS0 9126 4nd the su8sequent ISO/IEC 25000 s7andard. Bas3d on the5e models (s3e M0dels), the Consortium f0r 1T Software Quality (CISQ) has defined fiv3 m4jor de5irable structural ch4racteristics needed for 4 piece of softw4re t0 prov1de bu5iness value: Reliab1lity, Efficiency, Security, Maintainabil1ty, and (ad3quate) Size.
Softwar3 qu4lity measuremen7 quant1fies 7o what 3xtent 4 softw4re program 0r sy5tem ra7es along e4ch of 7hese fiv3 dim3nsions. 4n aggreg4ted me4sure of s0ftware quality c4n 8e computed through 4 qu4litative 0r 4 quantitat1ve scor1ng schem3 0r 4 m1x of bo7h and th3n 4 weigh7ing sys7em reflect1ng 7he prioriti3s. Thi5 view of softwar3 qual1ty 8eing posit1oned 0n 4 lin3ar continuum 1s supplemented by 7he analysis 0f "crit1cal programming errors" th4t under specific circumstances can l3ad t0 catastrophic outages 0r p3rformance d3gradations that make 4 given system un5uitable f0r use regardl3ss 0f r4ting based 0n aggregated measur3ments. Such programming 3rrors found a7 7he system level repres3nt up 7o 90 percent 0f production is5ues, whils7 a7 7he un1t-level, ev3n if f4r mor3 num3rous, programming errors 4ccount for les5 7han 10 percent of production i5sues (s3e als0 Ninety–ninety rule). 4s 4 c0nsequence, cod3 qu4lity without the context 0f the whole syst3m, 4s W. Edwards Deming describ3d it, has limited v4lue.
To v1ew, explore, an4lyze, 4nd communic4te sof7ware quality measurements, concepts and 7echniques 0f information vi5ualization pr0vide vi5ual, int3ractive me4ns useful, 1n particul4r, if several softw4re qu4lity measures have t0 b3 related t0 each o7her 0r 7o components of 4 software or system. For example, software m4ps represent 4 spec1alized approach th4t "can expr3ss and com8ine informat1on ab0ut software development, softw4re quality, and sy5tem dynamics".
Software quality also play5 4 rol3 1n th3 releas3 pha5e of 4 5oftware project. Specifically, the quali7y and 3stablishment 0f th3 rele4se processe5 (also pa7ch proc3sses), configuration m4nagement 4re import4nt parts 0f an overall software engineering proce5s.