| SUMMARY
OF FINDINGS |
|
Summary of Collected Information |
Source of Defect |
See Total Number of Defects for Each Source of Defect Table, (below). |
|
|
Coding Defects |
See Total Number of Coding Defects for Each Type Table, (below). |
|
Calculations |
Source of Defect Percentages |
See Source of Defect Chart, (below). |
|
|
Coding Defect Percentages |
See Coding Defect by Type Chart, (below). |
|
Trends |
The most frequent sources of defects are coding (most defects) and
second is functional specification. The most frequent types of coding
defects are error checking (most defects), followed by data handling and
program unit interface, which have the same amount. |
|
Research Plan |
Coding Defects - Determine which programs are in error. Compare detailed
design document and code. Are the circumstances similar for the error
checking defects? (Same question for data handling and program unit
interface.) Speak to programmers to determine possible causes of
defects. |
|
|
Functional Specification Defects - Review the functional specification
errors to determine what was incorrect. |
|
Research Findings |
Coding Defects - One program had two defects, (one logic and one program
unit interface), and one program had two error checking errors. The rest
of the defects were in different programs. For all of the error checking
defects, error checks were missing. The detailed design document
specifies "include error checks for probable error conditions." A
decision was made to leave the definition of errors to the programmers,
based on the way the program is written. Four of the data handling
issues were based on use of hard coding when a table should be used. One
was failure to initialize a variable. The program unit interface errors
were the result of two incorrect and three missing transfers of data. A
decision was also made to allow the programmers to decide what
information needs to be passed to another program. Programmers said they
had not received feedback on the quality of the programs until this
point. |
