2. Content
1 Challenges
2 Related work
3 Proposed approach
4 Application of the approach
5 Conclusion and future work
3. What is the problem?
Software importance increases a lot for industrial vendors
Product innovations more and more software based
Product assembly lines highly automated and permanently
measured
Software “assembly“ process is not software production
Improvements in software development are initiated
But difficult to prove objectively
Productivity figures needed for software measurement
Quantity is one required figure
How to measure?
4. This is necessary
# Requirement Description
R1 Figures on software output are concrete numbers
R2 Figures are objective (i.e., they are not based on questionnaires)
R3 Figures include the size of software and its development documentation
R4 The measurement of the figures is automated
R5 The figures are useable for trend analysis
R3
Industrial software produces a lot of documentation
Documents support life cycle management
Customer and certification authorities require thorough documentation
IEEE 610.12 definition of software: “computer programs, procedures
and possibly associated documentation.” [1]
[1] IEEE 610.12 - IEEE Standard Glossary of Software Engineering Terminology., Std., 1990
5. Content
1 Challenges
2 Related work
3 Proposed approach
4 Application of the approach
5 Conclusion and future work
6. Cool, the sliced V-Model [2]
Agility extension of
traditional V-model
Reduces Work in
Progress (WiP)
Documents are
container of Work
Items
Links between Work
Items to form “V”
[2] A. Deuter, “Slicing the V-model - Reduced effort, higher flexibility” in Proceedings of 8th International Conference on Global Software Engineering,
ICGSE’13, 2013, pp. 1–10.
7. Don‘t misunderstand the Work Item
Artefact created during
software development
Requirement
Specification
Test Case
Defect
Represent documentation
And can be also a
Task
8. Software productivity, hm?
Not defined in international Standards
Multiple approaches available. Petersen analyzed 586
articles with many different quantification approaches [2]
CMMI and SPICE require measurement, however no
required method given
Therefore, Industry vendors need to select
and adopt out of variety
We follow Sneed’s devils square [3]
[3] K. Petersen, “Measuring and predicting software productivity: A systematic map and review,” Inf. Softw. Technol., vol. 53, no. 4, pp. 317–343, Apr. 2011.
[4] H. M. Sneed, Software Management for Engineers, ser. Ellis Horwood series in information technology. Ellis Horwood Ltd, 1989.
9. We miss documentation measurement
Functional quantity measurement
All function point variants
Constructive quantity measurement
LOC, source-code analysis, churn
# Functional methods Constructive methods
R1 + +
R2 o +
R3 - -
R4 o +
R5 + +
10. Content
1 Challenges
2 Related work
3 Proposed approach
4 Application of the approach
5 Conclusion and future work
11. Churn is a measure for developers activity
Analyzing source-code
revisions
Get unified diff patch
for each file
Sum up per “module”
work item
Sum up per project
UPP, RevP,, FileP
12. Work Items show documentation size
Counting work items
linked with
requirements and
defect
RSize,DSize,WID,WIT
Sum up per project
Bottom-Up Analysis
13. Content
1 Challenges
2 Related work
3 Proposed approach
4 Application of the approach
5 Conclusion and future work
14. Who are we?
Phoenix Contact supplier of electrical and electronic
components for industrial applications.
13,000 employees, 51 owned sales companies and more than 30
sales partners worldwide.
Software activities are carried out in three locations Bad Pyrmont,
Lemgo (Germany), Ann Arbor (USA).
Sliced V-Model implemented
using Polarion ALM [5]
Reporting Tool implementing
above concept
[5] Polarion. http://www.polarion.com, 2014.
15. How far are we
Analyzing three software projects and one detailed
requirement:
Windows development (A)
Embedded development (B)
Reporting tool (C)
16. Yes, it works
Simple trends
Discussion with project and test manager on project A
Meaningful and helpful figures
17. Content
1 Challenges
2 Related work
3 Proposed approach
4 Application of the approach
5 Conclusion and future work
18. We are happy, …
R1: Concrete numbers
Number of work items, size of churn in kB
R2: Objective figures
Taken direct out of Polarion and Subversion database
R3: Software size including documentation
Work Items represent documentation
R4: Automated measurement
Stand-Alone reporting tool
R5: Trend analysis
Comparing Snapshots creates trends
19. …but much more to do
Work with these figures in daily routines
Possible detailing
Detailed analysis on work items, e.g. including text size
Analyzing changes in work items (“work item churn”).
Challenge.
Implement sliced V-model correctly in practice
Compare figures between projects
Integrate quantity figures in productivity measurements
Create figures for quality, costs and duration
Create meaningful relationships between these figures