FIT4004 Lecture Notes - Lecture 11: Software Maintenance, Code Coverage, Operand
25 Jun 2018
School
Department
Course
Professor
Lecture 11 - Quality Metrics
Why Measure?
To make good decisions about soware
●Predicons about product before we build it
●Assess product or process to decide what to do next
Types of Quality Metrics
●Funconality
●Reliability
●Usability
●Efficiency
●Maintainability
●Portability
Tesng Metrics
Mutaon Analysis
●Indicaon if the test suite is any good that just code coverage
●Involves automacally seeding a change into your source code
○Running the modified code with the test suite
○See whether the test suite report a fault
●Repeat this mulple mes
●Get a mutaon score from 0 to 1
●Use when its small soware, too expensive for larger soware
Design and Code Quality Metrics
Good for funconal correctness, maintainability and portability of code.
Proxy metrics
● Collected by stacally analysing source code and design arfacts
●Useful indicators of a number of things, not just the likelihood of funconal correctness
OO Design Metrics
●Look at structure of system
●Many only need class diagram, not the code
●Claimed to predict maintainability / testability
●Chidamber and Kemerer proposed them
Weighted Methods Per Class (WMC)
Sum of the complexies of all the methods:
Document Summary
P(cid:396)edi(cid:272)(cid:415)o(cid:374)s a(cid:271)out p(cid:396)odu(cid:272)t (cid:271)efo(cid:396)e (cid:449)e (cid:271)uild it. Assess p(cid:396)odu(cid:272)t o(cid:396) p(cid:396)o(cid:272)ess to de(cid:272)ide (cid:449)hat to do (cid:374)e(cid:454)t. I(cid:374)di(cid:272)a(cid:415)o(cid:374) if the test suite is a(cid:374)(cid:455) good that just (cid:272)ode (cid:272)o(cid:448)e(cid:396)age. I(cid:374)(cid:448)ol(cid:448)es auto(cid:373)a(cid:415)(cid:272)all(cid:455) seedi(cid:374)g a (cid:272)ha(cid:374)ge i(cid:374)to (cid:455)ou(cid:396) sou(cid:396)(cid:272)e (cid:272)ode. Ru(cid:374)(cid:374)i(cid:374)g the (cid:373)odi ed (cid:272)ode (cid:449)ith the test suite. Ee (cid:449)hethe(cid:396) the test suite (cid:396)epo(cid:396)t a fault. Get a (cid:373)uta(cid:415)o(cid:374) s(cid:272)o(cid:396)e f(cid:396)o(cid:373) (cid:1004) to (cid:1005) Use (cid:449)he(cid:374) its s(cid:373)all so(cid:332)(cid:449)a(cid:396)e, too e(cid:454)pe(cid:374)si(cid:448)e fo(cid:396) la(cid:396)ge(cid:396) so(cid:332)(cid:449)a(cid:396)e. Good fo(cid:396) fu(cid:374)(cid:272)(cid:415)o(cid:374)al (cid:272)o(cid:396)(cid:396)e(cid:272)t(cid:374)ess, (cid:373)ai(cid:374)tai(cid:374)a(cid:271)ilit(cid:455) a(cid:374)d po(cid:396)ta(cid:271)ilit(cid:455) of (cid:272)ode. Colle(cid:272)ted (cid:271)(cid:455) sta(cid:415)(cid:272)all(cid:455) a(cid:374)al(cid:455)si(cid:374)g sou(cid:396)(cid:272)e (cid:272)ode a(cid:374)d desig(cid:374) a(cid:396)(cid:415)fa(cid:272)ts. Useful i(cid:374)di(cid:272)ato(cid:396)s of a (cid:374)u(cid:373)(cid:271)e(cid:396) of thi(cid:374)gs, (cid:374)ot just the likelihood of fu(cid:374)(cid:272)(cid:415)o(cid:374)al (cid:272)o(cid:396)(cid:396)e(cid:272)t(cid:374)ess. Ma(cid:374)(cid:455) o(cid:374)l(cid:455) (cid:374)eed (cid:272)lass diag(cid:396)a(cid:373), (cid:374)ot the (cid:272)ode. U(cid:373) of the (cid:272)o(cid:373)ple(cid:454)i(cid:415)es of all the (cid:373)ethods: This (cid:449)ould p(cid:396)edi(cid:272)t the (cid:415)(cid:373)e a(cid:374)d e o(cid:396)t (cid:396)e(cid:395)ui(cid:396)ed to de(cid:448)elop a(cid:374)d (cid:373)ai(cid:374)tai(cid:374) the (cid:272)lass. La(cid:396)ge (cid:272)lasses (cid:449)ith (cid:374)u(cid:373)(cid:271)e(cid:396) of (cid:373)ethods (cid:373)o(cid:396)e likel(cid:455) to (cid:271)e appli(cid:272)a(cid:415)o(cid:374) spe(cid:272)i (cid:272) Nu(cid:373)(cid:271)e(cid:396) of t(cid:396)a(cid:374)si(cid:415)o(cid:374)s f(cid:396)o(cid:373) a (cid:396)oot (cid:374)ode.
