As per Sneed's study
in 1986, the number of bugs detected by running 90% "data coverage"
is twice as high as those detected by requiring 90% branch coverage
Statement and branch
coverage are found to be cost effective
Even when not
supported by automation, data flow testing has been found to be effective
Data flow testing
needs additional record keeping that is used to keep track of the statuses of
the variables, for which a computer is most effective
This will
significantly reduce the effort of data flow testing
Data flow testing
tools can be incorporated into compilers
Post a Comment
Post a Comment