Those slides were presented during the International Conference on Software Reuse (ICSR 2015).
This presentation highlights key aspect of our work on change impact analysis on highly variable products.
Call Us ≽ 9953322196 ≼ Call Girls In Lajpat Nagar (Delhi) |
Icsr2015
1. 1Challenge the future
Feature Change Impact on
Configurations of a Multi-product Line
Nicolas Dintzner, PhD student
Technical Univesity of Delft, Netherlands
Prof. Martin Pinzger
Software Engineering Research Group
University of Klagenfurt, Austria
Prof. Arie van Deursen
Software Engineering Research Group
Technical University of Delft, Netherlands
joint work with
Prof. Uira Kulesza
Federal University of Rio
Grande do Norte, Natal, Brazil
5. 5Challenge the future
Change impact of sub-systems evolution:
the video chain scenario
Video chain
Configuration 1
Interface:
Data/Video exchange
3rd
party
Equipment
Sub-system:
Video chain
Video chain
Configuration 2
Video chain
Configuration 3
Data/Video exchange
High resolution
Data/Video exchange
Low resolution
Video chain
Configuration 4
REQUIRES
IMPLEMENTS
IMPLEMENTS
6. 6Challenge the future
Requirements and Variability
with a catch
X-Ray system
Sub-system:
Data transfer
Sub-system:
Image processing
Sub-system:
Display
Interface:
Data/Video exchange
Interface
Video signal: input/output
Interface:
Monitors
3rd
party
Equipment
Interactions ?
Components ?
Existing configurations!
REQUIRES
7. 7Challenge the future
Targets
• Model the system using available information
• Build an approach to evaluate the impact of a
feature change
• Evaluate the approach using known evolution
scenarios
8. 8Challenge the future
Methodology
Sub-system specifications
3rd
party integration documents
System Model Impact computation technique
Evolution scenarios
3rd
party integration team
reviews
validates
output
9. 9Challenge the future
Modeling sub-systems
X-Ray system
Sub-system:
Data transfer
Sub-system:
Image processing
Sub-system:
Display
Interface:
Data/Video exchange
Interface
Video signal: input/output
Interface:
Monitors
3rd
party
Equipment
requires
Sub-system:
Data transfer
Sub-system:
Image processing
Sub-system:
(Display)
feature model: subsystem
components
configurations: known product
configurations
- C1: splitter; VGA port; …
- C2: repeater, Ethernet port,…
10. 10Challenge the future
Modeling interfaces
X-Ray system
Interface:
Data/Video exchange
Interface
Video signal: input/output
Interface:
Monitors
3rd
party
Equipment
requires
Sub-system:
Data transfer
Sub-system:
Image processing
Sub-system:
Display
Sub-system:
Data transfer
Sub-system:
Image processing
Sub-system:
(Display)
feature model: interface capabilities
configurations: 3rd
party
requirements
- C1: send video; import data; …
- C2: import video, send data,…
17. 17Challenge the future
Prototype: changes and simulation
Neo4JFM
Feature model 1Feature model 1 Feature model 2Feature model 2
Select feature to remove1
2 Load and update FMs
Configurations of FM-1 Configurations of FM-2
Feature model 1Feature model 1
Configurations of FM-1
Feature model 1
(EDITED)
Feature model 1
(EDITED)
3 Validate existing configurations
4 Check shared feature constraints
Configurations of FM-1 (EDITED)
Shared feature
constraints
FM 1 – FM 2
Shared feature constraints FM 1 – FM 2
5 Propagation to FM 2
Shared feature constraints
FM 1 – FM 2 (EDITED)
Configurations of FM-2
18. 18Challenge the future
Conclusions of the 3rd
party integration
case study
Unified modeling for interfaces and subsystems
Model composition and impact computation
Graph database for feature models
Naming convention
Feature notation usage
The interface uses only part of the interface (intro to “interface configuration”)
Question: given a highly variable system, after a change, can I still produce a system that satisfies the 3rd party product requirements.
They want to know if a change invalidates the integration of a 3rd party product.
Model based approach, solve this.
The interface uses only part of the interface (intro to “interface configuration”)
Interfaces : many interfaces of different natures (10 or so – physical, electrical, logical, …. )
Components : hundreds.
Shared hardware!
Existing work !!! SPL Approach for impact computation
----- Meeting Notes (03/12/14 16:29) -----
building the feature models
Shared hardware!
Cardinality in feature model – no go.
Alternative using model merging here – missing constraints between feature model, or over-arching structure for the feature models.