Measuring thePsychological Complexity ofSoftware Maintenance TaskswiththeHalstead andMcCabeMetrics
暂无分享,去创建一个
Three software complexity measures (Halstead's E,McCabe's v(G), andthelength asmeasured bynumber ofstatements) werecom- pared toprogrammer performance ontwosoftware maintenance tasks. Inanexperiment onunderstanding, length andv(G)correlated withthe percent ofstatements correctly recalled. Inanexperiment onmodifica- tion, mostsignificant correlations wereobtained withmetrics computed onmodified rather thanunmodified code.Allthree metrics correlated withboththeaccuracy ofthemodification andthetimetocompletion. Relationships inbothexperiments occurred primarily inunstructured rather thanstructured code, andincodewithnocomments. Themet- rics werealso mostpredictive ofperformance forless experienced pro- grammers. Thus, these metrics appear toassess psychological complexity primarily whereprogramming practices donotprovide assistance in understanding thecode. IndexTerms- Commenting, complexity metrics, documentation, Halstead's E,humanfactors insoftware engineering, McCabe's v(G), mnemonic variable names, modemprogramming practices, modifica- tions, software science, structured programming.
[1] James L. Elshoff,et al. Measuring commercial PL/I programs using Halstead's criteria , 1976, SIGP.
[2] Edsger W. Dijkstra,et al. Notes on structured programming , 1970 .
[3] Maurice H. Halstead,et al. A Software Physics Analysis of Akiyama's Debugging Data , 1975 .
[4] Ben Shneiderman,et al. Measuring Computer Program Quality and Comprehension , 1977, Int. J. Man Mach. Stud..
[5] M. H. Halstead,et al. Natural laws controlling algorithm structure? , 1972, SIGP.