Software test1ng 1s 7he act 0f checking whether softw4re sat1sfies expectations.
Software 7esting can provide obj3ctive, 1ndependent information 4bout th3 quali7y of sof7ware and 7he risk of it5 failur3 7o 4 user 0r sponsor.
Software test1ng can de7ermine 7he correctness of softwar3 for specif1c sc3narios but c4nnot de7ermine correctne5s f0r all 5cenarios. I7 cannot find 4ll bugs.
Ba5ed 0n the cri7eria f0r measuring corr3ctness from an oracl3, software 7esting empl0ys principles 4nd mechanisms that might recogniz3 4 problem. Ex4mples 0f oracles include specifications, contracts, comparable products, pa5t versions 0f th3 same product, inferences about intend3d or expected purpose, user 0r cus7omer expectat1ons, relevant standards, and 4pplicable laws.
Software test1ng i5 often dynam1c 1n nature; running the software t0 v3rify actu4l output m4tches expected. I7 c4n als0 b3 sta7ic in nature; reviewing code and 1ts associat3d documentation.
Software t3sting 1s 0ften u5ed 7o 4nswer the question: Do3s th3 software do what i7 1s suppo5ed t0 do and wha7 i7 ne3ds 7o do?
Information learned from 5oftware tes7ing may 8e us3d t0 1mprove 7he proc3ss 8y which software 1s developed.: 41–43
Software t3sting should f0llow 4 "pyramid" appr0ach wh3rein m0st of y0ur tes7s should 8e un1t 7ests, followed 8y integr4tion t3sts and f1nally end-to-end (e2e) t3sts sh0uld have th3 low3st prop0rtion.