Model-Driven Aspect Adaptation To Support Modular Software Evolution Ph.D. Defense Jing Zhang Committee Members: Dr. Jeff Gray (Academic Advisor) Dr. Barrett Bryant Dr. Aniruddha Gokhale Dr. Marjan Mernik Dr. Chengcui Zhang March 30th, 2009

2 Overview of Presentation Research Goals Motivation Challenges Approaches Evaluation Model-Driven Aspect Adaptation Aspect-Oriented Activity Modeling Bold Stroke Model Evolution Legacy Evolution Co-Evolution Inevitable Software Evolution Modular Evolution Modular Evolution ChangeabilityComprehensibility Independent Development INFM

3 First Law of Software Evolution “Software that is being used must be continually adapted or it becomes progressively less satisfactory.” -- Manny M. Lehman Manny M. Lehman, “Laws of Software Evolution Revisited,” In Proceedings of the 5th European Workshop on Software Process Technology (EWSPT), Nancy, France, October 1996, pages 108–124.

4 Challenges in Software Evolution: 1. Model Evolution ∆ M : The changes made to the models As the size of system models grows, techniques and tools are needed to automate complex change evolution

5 Challenges in Software Evolution: 2. Legacy Evolution ∆ S : The changes made to the legacy source A small modification in the high-level requirements may trigger drastic manual changes in large portions of the low-level legacy source code

6 Challenges in Software Evolution: 3. Model/Code Co-Evolution ∆ M : The changes made to the models ∆ S : The changes reflected in the source code Conformity between the models and underlying source code needs to be established in order to maintain the fidelity of the whole system

7 Criteria for Software Evolution Modularization  Dividing a complex system into a set of manageable building blocks, or modules Modular software evolution  Changeability Individual modules should be able to change without radical impact to the rest of the system  Comprehensibility High-level evolutionary requirements need to be well-understood and thus systematically translated into the low-level change adaptation  Independent Development Evolution tasks need to be divided into individual modules that can be implemented and maintained by different groups

8 Research Objectives Model Evolution Model/Code Co-Evolution Model/Code Co-Evolution Legacy Evolution Legacy Evolution ChangeabilityComprehensibility Independent Development Aspect- Orientation Model-Driven Aspect Adaptation

9 Thesis Statement This dissertation research is devoted to approaches uniting Model-Driven Engineering (MDE) with Aspect- Oriented Software Development (AOSD) techniques to support modular software evolution. The ultimate goal is to improve changeability, comprehensibility and independent development of the evolutionary task implementation.  First, the Aspect-Oriented Activity Modeling (AOAM) approach is investigated to support evolution of UML activity models.  Second, the Model-Driven Aspect Adaptation (MDAA) approach is investigated to facilitate model/code co-evolution and legacy evolution.

10 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

11 Two Types of Software Restructuring Horizontal transformation  Transformation within the same level of abstraction  Model Transformation, Aspect-Oriented Modeling (AOM), Program Transformation, Aspect-Oriented Programming (AOP) Vertical translation  Translation, or synthesis, between layers of abstraction  Reverse Engineering, Model-Driven Engineering (MDE) ComputePosition C++ ComputePosition with Locking C++ NavDisplay C++

12 Model-Driven Engineering OMG UML-based generic modeling  Rational Rose  Telelogic TAU  … Domain-specific modeling (DSM)  GME  Microsoft software factories  …

13 Unified Modeling Language (UML) Structural modeling  Class diagram  Component diagram  Deployment diagram Behavioral modeling  State machine diagram  Activity diagram  Use case diagram  Interaction diagram

14 Telelogic TAU: A UML-based MDE Tool Design Code Generation Testing Simulation

15 Model Interpretation Model Interpreters Models Modeling Environment Application Domain App 1 App 2 App 3 Application Evolution Environment Evolution Metaprogramming Interface Formal Specifications Meta-Level Translation Model Builder Domain-Specific Modeling (DSM) in Generic Modeling Environment (GME) MetamodelMetamodel ModelModel InterpreterInterpreter void CComponent::InvokeEx(CBuilder &buil der,CBuilderObject *focus, CBui lderObjectList &selected, long param) { CString DMSRoot = ""; DMSRoot = SelectFolder("Please Select DMS Root Folder:"); if (DMSRoot != "") { DMSRulePath = DMSRoot + RULESPATH + "Rules\\"; MSRuleApplierPath = DMSRoot + RULESPATH + "RuleApplier\\"; AfxMessageBox("DMSRulePath = " + DMSRulePath, MB_OK); CString OEPRoot = ""; OEPRoot = SelectFolder("Please Selec DEFINE INTERPRET

16 Crosscutting Concerns Problems  Reusability  Maintainability  Evolvability Profiling in Apache Tomcat Server

17 Aspect-Oriented Software Development (AOSD) Advanced separation of concerns  Modularize crosscutting concerns Aspect  Encapsulates each crosscutting concern in an individual module Aspect weaver  Composes aspects with the base modules

18 AOSD Terminology Join Point  A point of interest through which two or more concerns may be composed Pointcut  A collection of join points Advice  The adaptation to be applied at each join point specified in a pointcut Aspect applicability Aspect functionality Where What Aspect Pointcut Advice

19 AOSD Technologies Aspect-Oriented Programming (AOP)  AspectJ The most widely used AOP language Tailored for Java Aspect-Oriented Modeling (AOM)  Constraint-Specification Aspect Weaver (C-SAW) Aspect model weaver for domain-specific models Embedded Constraint Language (ECL) to support model weaving A plug-in for GME  Motorola WEAVR Aspect model weaver for UML state machines A plug-in for Telelogic TAU

20 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

21 Aspect-Oriented Activity Modeling

22 Order Processing Activity Model Operation Action Send Signal Action Receive Event Action Fork

23 Aspect-Activity Metamodel (Profile)

24 Tracing Aspect, Pointcuts and Advice

25 Order Processing Model with Tracing Aspect

26 Pointcut Composition in AOAM Operators  and (&&)  or (||)  not (!)  cflow  cflowbelow  within

27 Advice Composition in AOAM Operators  binds  follows Advice are ordered based on the precedence relationships (e.g., >)  Advice1 has precedence over Advice2 at the shared join point  Interference between the advice is reduced by declaring the advice precedence explicitly Advice1 Advice2 Shared Join Point

28 Aspect Composition in AOAM Operators  crosscuts  follows  hidden_by  dependent_on Interference between the aspects is reduced by declaring the aspect precedence explicitly TracingAspect follows ExceptionAspect  ExceptionAspect has higher precedence than TracingAspect LoggingAspect is hidden by TracingAspect  The presence of TracingAspect inactivates LoggingAspect LoggingAspect is dependent on EncryptionAspect  LoggingAspect can only be applied in the presence of the EncryptionAspect Exception Aspect Tracing Aspect Logging Aspect Encryption Aspect

29 Summary of AOAM Contributions  AOAM supports modular evolution in activity modeling Evolutionary requirements are encapsulated in aspect models Aspect models allow specific concerns to be maintained, comprehended and developed independently Aspect models are woven into the base model via the AOAM weaver  AOAM allows precedence relationships to be specified at the modeling level to prevent undesirable interferences between aspects  AOAM weaver is implemented in Telelogic TAU, as an extension of the Motorola state machine weaver

30 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

31 Model-Driven Aspect Adaptation Framework Meta- metamodel Metamodel Source Code Conforms To Model ExtractorModel Composer Enhanced Model Enhanced Code Program Composer Aspect Code Generator Aspect Code Aspect Model Aspect Metamodel Conforms To Simulation/Testing

32 Roles in MDAA Process

33 MDAA Component: Model Extractor DSM Metamodel UML Metamodel

34 MDAA Component: Model Composer AOM Weaver  C-SAW  AOAM Weaver

35 MDAA Component: Aspect Code Generator

36 MDAA Component: Program Composer Program transformation engine  Design Maintenance System (DMS)  ASF + SDF …… AOP weaver  AspectJ  AspectC++ ……

37 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

38 DSM-Based MDAA Instantiation Evolutionary requirements are captured as higher level policy strategies and weaved into models Ensures causal connection between model changes and the underlying source code of the legacy system Large-scale adaptation across multiple source files that are driven by minimal changes to the model properties Domain experts are shielded from lower-level transformation rules

39 Metamodel Composer Why metamodel composer?  C-SAW provides no support for transformation BETWEEN two different metamodels  Need a means to compose base metamodel and aspect metamodel Metamodel Composer

40 Experimental Case Study: Bold Stroke Product Line Background context  Mission-control software for Boeing military aircraft under development since 1995  CORBA event-based systems  Thousands of components implemented in over three million lines of C++ code  Embedded System Modeling Language (ESML) Key challenges  Difficult to evolve the underlying source representation to address new requirements  Impossible to determine, a priori, all of the future adaptation requests

41 ESML Base Metamodel

42 Case Study: A Black Box Data Recorder Requirement: Record the state information of the aircraft according to polices defined in a model Under different stages of development, a concern may need to be adjusted based on different contexts, e.g., testing on the ground vs. in-flight recording Ability to rapidly explore design alternatives that represent different policies

43 Composed ESML Metamodel with Logging Aspect

44 Apply C-SAW to Insert Logging Aspect to ESML Base Model defines Start, logDataAtoms, AddLog; strategy logDataAtoms() { atoms()->select(a | a.kindOf() == "Data")->AddLog(); } strategy AddLog() { declare parentModel : model; declare dataAtom, logAtom : atom; dataAtom := self; parentModel := parent(); logAtom := parentModel.addAtom("Log", "LogOnMethodExit"); logAtom.setAttribute("Kind", "On Method Exit"); logAtom.setAttribute("MethodList", "Update"); parentModel.addConnection("AddLog", logAtom, dataAtom); } aspect Start() { rootFolder().findFolder("ComponentTypes").models(). select(m|m.name().endWith("Impl"))->logDataAtoms(); }

45 ESML Component with Logging Aspect

46 Generated DMS Transformation Rules to Insert “LogOnMethodExit” Aspect default base domain Cpp~VisualCpp6. pattern LogStmt(): statement = "log.add(\"data1_=\" + data1_);". pattern LogOnMethodExitAspect(s: statement_seq): statement_seq = "\s \LogStmt\(\)". pattern JoinPoint(id:identifier): qualified_id = "\id :: Update". rule insert_log_on_method_exit(id:identifier, s:statement_seq, p:parameter_declaration_clause): function_definition -> function_definition = "void \JoinPoint\(\id\)(\p) {\s} " -> "void \JoinPoint\(\id\)(\p) {\LogOnMethodExitAspect\(\s\)}" if ~[modsList:statement_seq. s matches "\:statement_seq \LogOnMethodExitAspect\(\modsList\)"]. public ruleset applyrules={insert_log_on_method_exit}.

47 Transformed C++ Source Code 1 unsigned int BM__ClosedEDComponentImpl::getData1_ () const 2 { UM__GUARD_INTERNAL_REGION; 6 BM__ComponentInstrumentation::ReceiveDirectCall(GetId(), "GetData1"); return data1_; 10 } void BM__ClosedEDComponentImpl::Update (const UUEventSet& events) 13 { UM__GUARD_EXTERNAL_REGION(GetExternalPushLock()); 17 BM__ComponentInstrumentation::EventConsumer(GetId(), "Update", events); 18 unsigned int tempData1 = GetId().GetGroupId(); 19 unsigned int tempData2 = GetId().GetItemId(); //*** REMOVED: code for implementing Real-time Event Channel data1_ = tempData1; //*** REMOVED: actual variable names (proprietary) 25 data2_ = tempData2; } log.add("data1_=" + data1_); Log on getData1_() method entry Log on reading data1_ Log on Update() method entry Log on writing data1_ Log on Update() method exit

48 Experimental Results Manual Changes (No. Of Places) Aspect Model Interpreter (LOC) ECL (LOC) Generated RSL (LOC) Concurrency50 (in 25 different cpp files) Logging200 (in 25 different cpp files) Assertion100 (in 25 different cpp files)

49 MDAA Interpreters for Different Aspect Models

50 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

51 Activity-Based MDAA Instantiation

52 Experimental Case Study: Intelligence Network Fault Management (INFM) Background context  A fault management software developed by Motorola Labs for CDMA cellular network  Alarm correlation by pattern discovery algorithms  Implemented in over 50K lines of Java code Key challenges  Increasing computational complexity would cause various software failures (e.g., timeout failure and pattern failure)  Different strategies are needed to resolve software failures  Failure management strategies are subject to change with evolving system requirements

53 INFM Alarm Correlation Activity Base Model

54 Case Study: A Pattern Failure Handler Aspect

55 Composed INFM Model by AOAM Weaver

56 Generated AspectJ Code for the Pattern Failure Handler Aspect aspect PatternFailureHandlerAspect { int N = 10; int counter = 0; boolean hasFailure; Param deltaParam; pointcut pct() : call(ExecuteAlgo(...)); around() : pct() { while (counter < N) { proceed(); hasFailure = IdentifyPatternFailure(); if (hasFailure) { hasFailure = RecheckPatterns(); if (hasFailure) { deltaParam = AnalyzePatternFailure(); ReconfigureAlgo(deltaParam); counter ++; } else { break; } } else { break; } }

57 Summary of MDAA MDAA provides modular evolution for legacy systems by uniting MDE and AOSD technologies  Evolutionary changes are encapsulated in high-level abstract aspect models  Aspect models allow specific evolutionary requirements to be maintained, comprehended and developed independently  Aspect models are translated into the low-level aspect code by aspect code generators  The legacy system is transformed by weaving the generated aspect code into the base source code

58 Comparison of DSM-based and UML-based MDAA Instantiation DSM-based MDAAUML-based MDAA Base MetamodelDomain-specificDomain-generic ModelClose to the domain concepts Close to the implementation concepts Aspect MetamodelMultiple for one base metamodel Only one for one base metamodel Aspect Code Generator Multiple for one base metamodel Only one for one base metamodel

59 Outline Motivation and Challenges Research Objectives Background Aspect-Oriented Activity Modeling Model-Driven Aspect Adaptation  DSM-Based MDAA  UML Activity-Based MDAA Future Work and Conclusion

60 Future Work AOM  More types of join points support in AOAM  Aspect mining and refactoring at the model level Extend initial experimental work on aspect identification to support aspect extraction and refactoring  Aspect interference analysis Extend preference-based approach for reducing aspect interference Perform formal evaluation and analysis for detecting aspect interference MDAA  System validation through model simulation  Version control support  MDA (Model-Driven Architecture) and ADM (Architecture-Driven Modernization) alignment

61 Conclusion Software evolution challenges  Activity model evolution Solution: AOAM weaver  Model/Code co-evolution, legacy evolution Solution: MDAA framework Contribution  Modular software evolution by combining MDE and AOSD Changeability – aspect models are able to change without radical impact to the rest of the system Comprehensibility – evolutionary requirements are captured in high- level aspect models and systematically translated into the low-level change adaptation Independent Development – evolution tasks are divided into individual aspect modules that can be developed and maintained by different groups

62 Representative Publications Book chapters (3):  Jeff Gray, Sandeep Neema, Jing Zhang, Yuehua Lin, Ted Bapty, Aniruddha Gokhale, and Douglas C. Schmidt, “Concern Separation for Adaptive QoS Modeling in Distributed Real-Time Embedded Systems,” in Behavioral Modeling for Embedded Systems and Technologies: Applications for Design and Implementation, Idea Group,  Jing Zhang, Yuehua Lin, and Jeff Gray, "Generic and Domain- Specific Model Refactoring using a Model Transformation Engine," Model-driven Software Development - Research and Practice in Software Engineering, Springer, ISBN: X, 2005, Chapter 9, pp  Yuehua Lin, Jing Zhang, and Jeff Gray, "A Framework for Testing Model Transformations," Model-driven Software Development - Research and Practice in Software Engineering, Springer, ISBN: X, 2005, Chapter 10, pp

63 Representative Publications (Cont’) Journal papers (7):  Suman Roychoudhury, Jeff Gray, Jing Zhang, Purushotham Bangalore, and Anthony Skjellum, “Modularizing Scientific Libraries With Aspect-Oriented And Generative Programming Techniques,” Acta Electrotechnica et Informatica Journal,  Yuehua Lin, Jeff Gray, Jing Zhang, Steve Nordstrom, Aniruddha Gokhale, and Sandeep Neema, “Replicators: Transformations to Address Model Scalability,” Software: Practice and Experience, vol. 38, no. 14, November 2008, pp  Jing Zhang, Jeff Gray, Yuehua Lin, and Robert Tairas, “Aspect Mining from a Modeling Perspective,” International Journal of Computer Applications in Technology, Volume 31, Number 1-2, March 2008, pp  Faizan Javed, Marjan Mernik, Jeff Gray, Jing Zhang, Barrett Bryant, and Suman Roychoudhury, “Using A Program Transformation Engine to Infer Types in A Metamodel Recovery System,” Acta Electrotechnica et Informatica, vol. 8, no. 1, January-March 2008, pp  Jing Zhang, Thomas Cottenier, Aswin van den Berg, and Jeff Gray, “Aspect Composition in the Motorola Aspect-Oriented Modeling Weaver,” JOT special issue on AOM, August 2007, pp  Krishnakumar Balasubramanian, Aniruddha Gokhale, Yuehua Lin, Jing Zhang, and Jeff Gray, “Weaving Deployment Aspects into Domain-Specific Models,” International Journal on Software Engineering and Knowledge Engineering, vol. 16, no. 3, June 2006, pp  Jeff Gray, Yuehua Lin and Jing Zhang, “Automating Change Evolution in Model- Driven Engineering,” IEEE Computer (Special Issue on Model-Driven Engineering), vol. 39, no. 2, February 2006, pp

64 Representative Publications (Cont’) Conference papers (11):  Michael Jiang, Jing Zhang, Hong Zhao, and Yuanyuan Zhou, “Enhancing Software Product Line Maintenance with Source Code Mining,” International Conference on Wireless Algorithms, Systems and Applications (WASA), Dallas, TX, October 2008, pp  Yan Liu, Jing Zhang, Xin Meng, and John Strassner, “Sequential Proximity-based Clustering for Telecommunication Network Alarm Correlation,” The Fifth International Symposium on Neural Networks (ISNN), Beijing, China, September 2008, pp  Michael Jiang, Jing Zhang, Hong Zhao, and Yuanyuan Zhou, “Maintaining Software Product Lines – an Industrial Practice,” International Conference on Software Maintenance (ICSM), Beijing, China, September 2008, pp  John Strassner, Srini Samudrala, Greg Cox, Yan Liu, Michael Jiang, Jing Zhang, Sven van der Meer, Micheal Ó Foghlu and Willie Donnellu, “The Design of a New Context-Aware Policy Model for Autonomic Networking,” The 5th IEEE International Conference on Autonomic Computing (ICAC), Chicago, IL, June 2008, pp  Yan Liu, Jing Zhang, Michael Jiang, Dave Raymer, and John Strassner, “A Model-based Approach to Adding Autonomic Capabilities to Network Fault Management System,” IEEE/IFIP Network Operations and Management Symposium (NOMS 2008), Salvador, Brazil, April 2008, pp  Yan Liu, Jing Zhang, Michael Jiang, Dave Raymer, and John Strassner, “A Case Study: A Model-Based Approach to Retrofit a Network Fault Management System with Self-Healing Functionality,” 5th IEEE International Conference on Engineering of Autonomic and Autonomous Systems (EASe), Belfast, Northern Ireland, March/April 2008, pp  Jing Zhang, Yan Liu, Michael Jiang and John Strassner, “An Aspect-Oriented Approach to Handling Crosscutting Concerns in Activity Modeling,” IAENG International Conference on Software Engineering, Hong Kong, March 2008, pp  Best Paper Award: Jeff Gray, Yuehua Lin, Jing Zhang, Steve Nordstrom, Aniruddha Gokhale, Sandeep Neema, and Swapna Gokhale, “Replicators: Transformations to Address Model Scalability,” Model Driven Engineering Languages and Systems (MoDELS) (formerly the UML series of conferences), Springer-Verlag LNCS 3713, Montego Bay, Jamaica, October 2005, pp  Jeff Gray, Jing Zhang, Yuehua Lin, Hui Wu, Suman Roychoudhury, Rajesh Sudarsan, Sandeep Neema, Feng Shi, and Ted Bapty, “Model-Driven Program Transformation of a Large Avionics Application,” Generative Programming and Component Engineering (GPCE 2004), Springer-Verlag LNCS, Vancouver, BC, October 2004, pp  Song Zhou, Chuanxi Xu, Hui Wu, Jing Zhang, Yuehua Lin, Juanqin Wang, Jeff Gray, and Barrett Bryant, “E-R Modeler: A Database Modeling Toolkit for Eclipse,” 42nd Annual ACM SE Conference, Huntsville, AL, April 2004, pp  Suman Roychoudhury, Jeff Gray, Hui Wu, Jing Zhang, and Yuehua Lin, “A Comparative Analysis of Meta- programming and Aspect-Orientation,” 41st Annual ACM SE Conference, Savannah, GA, March 7-8, 2003, pp

65 Representative Publications (Cont’) Workshop papers (10):  Yan Liu, Jing Zhang, and John Strassner, “Model-Driven Adaptive Self-Healing for Autonomic Computing,” The 3rd IEEE International Workshop on Modeling Autonomic Communication Environments (MACE 2008), Samos Island, Greece, September 22-26,  John Strassner, Yan Liu and Jing Zhang, “A Context-Aware Policy Model to Support Autonomic Networking,” The First IEEE International Workshop on Model-Driven Development of Autonomic Systems (MDDAS) - A Workshop held at COMPSAC, Turku, Finland, July-August  Michael Jiang, Jing Zhang, David Raymer and John Strassner, “A Modeling Framework for Self-Healing Software Systems,” - A Workshop held at MoDELS Conference, Nashville, TN, October  Jing Zhang, Thomas Cottenier, Aswin van den Berg, and Jeff Gray, “Aspect Interference and Composition in the Motorola Aspect-Oriented Modeling Weaver,” MoDELS Workshop on Aspect-Oriented Modeling, Genova, Italy, October  Arundhati Kogekar, Dimple Kaul, Aniruddha Gokhale, Paul Vandal, Upsorn Praphamontripong, Swapna Gokhale, Jing Zhang, Yuehua Lin, Jeff Gray, “Model-driven Generative Techniques for Scalable Performabality Analysis of Distributed Systems,” IPDPS Workshop on Next Generation Systems, Rhodes Island, Greece, April  Jing Zhang, Jeff Gray, and Yuehua Lin, “A Model-Driven Approach to Enforce Crosscutting Assertion Checking,” 27th ICSE Workshop on the Modeling and Analysis of Concerns in Software (MACS), St. Louis, MO, May  Jing Zhang, Jeff Gray, and Yuehua Lin, “A Generative Approach to Model Interpreter Evolution,” 4th OOPSLA Workshop on Domain-Specific Modeling, Vancouver, BC, Canada, October 2004, pp  Yuehua Lin, Jing Zhang, and Jeff Gray, “Model Comparison: A Key Challenge for Transformation Testing and Version Control in Model Driven Software Development,” OOPSLA Workshop on Best Practices for Model-Driven Software Development, Vancouver, BC, Canada, October  Jing Zhang and Jeff Gray, “Legacy System Evolution through Model-Driven Program Transformation,” 8th IEEE International Enterprise Distributed Object Computing Conference (EDOC 2004) Workshop on Model- Driven Evolution of Legacy Systems (MELS), Monterey, CA, September  Jeff Gray, Yuehua Lin, and Jing Zhang, “Aspect Model Weavers: Levels of Supported Independence,” Middleware 2003: Workshop on Model-driven Approaches to Middleware Applications Development, Rio de Janeiro, Brazil, June 2003.

66 Publication, Software and Video demos available at:

67 Backup

68 Application Evolution Evolution Scenario 1: Application Evolution Model Interpretatio n Model Interpreters Models Modeling Environment Application Domain App 1 App 2 App 3 Environment Evolution Metaprogramming Interface Formal Specifications Meta-Level Translation Model Builder  Changes are made to the domain models  Model interpretation re-synthesizes the whole application system Application Evolution Complete regeneration is not always a plausible solution when applied to existing large legacy systems

69 Environment Evolution Evolution Scenario 2: Environment Evolution Environment Evolution  Changes are made to the metamodel  Model migration to support domain model evolution Model Interpretatio n Model Interpreters Models Modeling Environment Application Domain App 1 App 2 App 3 Metaprogramming Interface Formal Specifications Meta-Level Translation Model Builder Application Evolution Advanced mechanism is needed to support crosscutting evolutionary concerns

70 Comparison of DSM to Programming Language and Database Definition Domain-Specific Modeling Programming Language Definition Database Schema Definition Schema definition Metamodel for a specific domain (e.g., Petri Net) Grammar for a specific language (e.g., Java) Table, constraint, and stored procedure definitions for a specific domain (e.g., payroll database) Schema instance Domain model (e.g., Petri Net model of a teller machine) A program written in a specific language Intension of a database at a specific instance in time (e.g., the June 2008 payroll instance) Schema execution Model interpreter Language compiler/interpreter Transactions and behavior of stored procedures in an executing application