Quick fix for domain-specific modeling languages using the VIATRA framework (http://viatra.inf.mit.bme.hu/) presented at the VL/HCC 2011 conference (http://www.cs.cmu.edu/~vlhcc2011/)

1. Quick fix generation for DSMLs
Ábel Hegedüs, Ákos Horváth, István Ráth, Moisés Castelo Branco*, Dániel Varró
Budapest University of Technology and Economics
*University of Waterloo
VL/HCC 2011
Pittsburgh, 2011.09.19.
Budapest University of Technology and Economics
Department of Measurement and Information Systems

2. Source code editing and fixing errors
Visual Studio Smart Compile Auto Correction
Netbeans Editor Hints
Eclipse Quick Fix
2

3. Development support in programming languages
 Economy class Examples:
o Syntax-highlighting, Custom DSL
lightweight validation
 Business class
o Autocompletion, Standard DSL (e.g. SQL)
semantic validation
 First-class
o Quick fixes, refactoring Java, C#, PHP
Eclipse-like IDEs
3

4. Development support in modeling languages
 Economy class Examples:
o Syntax-driven Traditional DSM,
editing COTS modeling tools
 Business class
o Validation UML tools
Modern DSM
 First-class Need for generic approach
o Quick fixes, for fix generation:
refactoring • Language-independent
?
• Customisable
• Efficient
4

5. Domain Specific Modeling Languages
 Ubiquitous in software development
o Requirement specification
o System analysis and verification
o Code generation
o Business processes
o Standard DSMLs e.g.
• Matlab Simulink
• UML profiles
• BPEL, BPMN
o Custom DSMLs
 Sophisticated DSML
development tools
o Eclipse Modeling Framework
o Microsoft DSL Tools
o MetaEdit
5

6. Domain-specific constraints
 Syntax-driven editing is not enough!
o Design errors, company policies, best practices
 Additional inconsistency rules
o Well-formedness constraints Syntactically correct
o Structural correctness process
Violates basic process
modeling practices
6

7. Example: BPMN process validation
Standard How to fix these
Eclipse BPMN tool inconsistencies?
7

8. Quick fix generation for DSMLs
 Requirements Graph patterns and
o Domain-specific pattern matching
• Inconsistency rules
• Editing operations
o Problem-specific
End event start of edge
 Output
 Essence of the approach Non-messaging event
o Design Space Exploration receiving message
8

9. Quick fix generation for DSMLs
 Requirements Graph patterns and
o Domain-specific pattern matching
• Inconsistency rules
• Editing operations Graph
o Problem-specific transformation
• Input model
• Violation context
 Output Post
o Sequence of applicable editing operations
o Eliminate violations in the selected context Pre
Quick fix
 Essence of the approach Modify target of sequence edge
o Design Space Exploration
9

10. Design Space Exploration (DSE)
Application of a overview
High-level syntax-
driven editing operation Initial (inconsistent) model
Operation Violations identified
by inconsistency rules
Modified model
Termination condition:
Violations fixed on Partially corrected • number of solutions
selected context • time limit
model
• depth
10

11. Design Space Exploration (DSE)
 High-level overview
Operation Initial (inconsistent) model
Exploration guided by
sequence =Operation • heuristics
quick fix • already visited states
Modified model
Efficient application
of operations
• Graph
Efficient
Transformation
backtracking of
Efficient solution
operations
detection
Partially corrected • violations evaluated
model incrementally
11

12. Quick demonstration: BPMN quick fix
12

13. Presentation of quick fixes
Sequence of
List of fixes Operation name Context
operations
13

14. Case studies and evaluation
 Main evaluation criteria
o „On-the-fly” generation of quick fixes
o Representative set of
• Inconsistency rules
• Editing operations
• Input models and violations
 Process size and complexity representative to
typical BPMN processes and simpler DSML models
 Real-life processes from the financial domain
o Centralized customer registering
o Corporate procurement
14

15. Evaluation results
Logarithmic scale
1st: 5ms Total:
2nd: 13ms 1 sec vs 160 sec részletek
egyéb
1 sec vs 160 sec
3rd: 271ms
Memory:
50-200 MB
Visited states:
50-40000
 „On-the-fly” quick fix generation works in many cases
 Finding multiple solutions can increase runtime significantly
15

16. Evaluation results
 Identified significant factors
o Number of local violations
o Operation complexity
o Solution density
 Usable during editing for fixing local errors
o 12 inconsistency rules
o 16 editing operations (create,modify,delete)
o 1-3 local violation, at most 2 step fix
16

17. Future work
 Knowledgebase-driven DSE
o User-aided fix generation
o Learn applied fixes for future use
• Store quick fix as a special editing operation
 Improved search strategy
o Partial order reduction
o Better heuristics
 Resumable search
 Streamlined integration with EMF tools
o Ease adaptation to third-party software
17

18. Summary
 Novel approach
o for an important language engineering problem
o applicable to graph-based modeling languages
 Problem formalized and solved efficiently
o using graph transformations, incremental pattern matching
and design space exploration
 Prototype implementation
o Evaluated on BPMN
o Also presented at the Demo session (tonight)
 We would like to thank the Bank of the Northeast of Brazil (Banco
do Nordeste – BNB) for providing the case studies.
Thank you!
Contact: hegedusa@mit.bme.hu
18

19. References
VIATRA2 Model EMF IncQuery
Transformation Framework
viatra.inf.mit.bme.hu viatra.inf.mit.bme.hu/incquery
Fault Tolerant Systems Research Group
Budapest University of Technology and Economics
inf.mit.bme.hu
Generative Software Development Lab
University of Waterloo
gsd.uwaterloo.ca