1 UDME Unit Development, Measurement and Efforts A programmer‘s experience What is possible?
2 U nit D evelopment, M easurement and E fforts Agenda q Problem q Process q Steps and Tests q System q Efforts q Errors q Measures q Conclusion
3 U nit D evelopment, M easurement and E fforts The Problem Too expensive... Too difficult... Too long... Too... Too...
4 U nit D evelopment, M easurement and E fforts Process
5 U nit D evelopment, M easurement and E fforts Basics for Start: Details
6 U nit D evelopment, M easurement and E fforts Standard Test: Details
7 U nit D evelopment, M easurement and E fforts Measured Quality: Details
8 U nit D evelopment, M easurement and E fforts Algorithm
9 U nit D evelopment, M easurement and E fforts Asserts assert ( tabel space ) assert ( overflow ) assert ( compiler restriction ) assert ( declaration of table limits ) assert ( underflow )
10 U nit D evelopment, M easurement and E fforts Standard Test Case Tree (in Parts) low = high + - more than one x i a Asserts are tested separately. i = low no x i a x i > a x i = a !
11 U nit D evelopment, M easurement and E fforts Planning the System Per Level Order of importance Conditions sine qua non
12 U nit D evelopment, M easurement and E fforts Implementing the System Hardest first Bottom up Complete sub tree Independent first Condition first Sub tree first
13 U nit D evelopment, M easurement and E fforts Selected Location Parameter Data collected: # of modules: 168 Effort in minutes Size in editor lines Maintenance index no dimension
14 U nit D evelopment, M easurement and E fforts Particular Efforts
15 U nit D evelopment, M easurement and E fforts Efforts Groups
16 U nit D evelopment, M easurement and E fforts Errors Benefit dynamic measurement 5:1
17 U nit D evelopment, M easurement and E fforts Maintenance Measuring Measure Tool Formula
18 U nit D evelopment, M easurement and E fforts Comment and MI
19 U nit D evelopment, M easurement and E fforts Regions of Maintenance More than 95 % of the modules have at least a fair maintainability index. Benefit of static measurement
20 Unit Development, Measurement and Efforts Effort and Poor Maintenance Within poor region HP (n=5 %=2.98) Effort Type- --Group--- Percent mean part mean all- Basic Std. Test Measurement Sum Effort
21 U nit D evelopment, M easurement and E fforts Effort and Fair Maintenance Within fair region HP (n=28 %=16.67) Effort Type- --Group--- Percent mean part mean all- Basic Std. Test Measurement Sum Effort
22 U nit D evelopment, M easurement and E fforts Effort and Excellent Maintenance Within excellent region HP (n=135 %=80.36) Effort Type- --Group--- Percent mean part mean all- Basic Std. Test Measurement Sum Effort Honi soit qui mal y pense!
23 U nit D evelopment, M easurement and E fforts It is possible... To develop programs so that Measuring requires less than 5 % of the effort, Test reach a code coverage of 100% LCSAJs within less than 2 % of the effort, Source codes fulfils the high level quality requirements without any additional effort for 98 % of the modules, An excellent maintainability is achieved with less effort per module than the mean value for more than 80 % of the modules, A poor maintainability appears for less than 3 % of the modules and Quality benefits are earned during the development process.
24 U nit D evelopment, M easurement and E fforts