Automated Analysis and Code
Generation for Domain-Specific Models
George Edwards
Center for Systems and Software Engineering
University of Southern California
gedwards@usc.edu
Presentation Outline
• Background: Domain-Specific Languages and Model-Driven
Engineering
• Research Challenge: Code Generation and Analysis Using
Domain-Specific Models
• Promising Solution: Model Interpreter Frameworks
• XTEAM: Extensible Toolchain for Evaluation of Architectural
Models
Domain-Specific Languages
• Software and systems modeling languages
that are customized for a particular context
• Concisely and intuitively express the essential
features of system designs
– No missing or extraneous features
– Capture common, reusable patterns
– Enforce of architectural constraints
– Use symbols and names native to the domain
• Easily modified, evolved, and composed
DSL Examples
• Specialized types,
attributes, and
relationships
• Constraints on how
types may be used
• Customized symbols
and multiple model
views
DSL Specification
Common DSL Uses
• Architecture DSLs
• Architectural styles
• Reference architectures
• Product line architectures
• Platform DSLs
• Middleware configuration
• Deployment descriptors
• Code generation for
application frameworks
• Analysis DSLs
• Simulation
• Model checking
• Quantification of quality
properities
Why Use DSLs?
• Heterogeneity of platforms and technologies
– Examples: multiple types of networks, data models, middleware
– Result: A “one-size-fits-all” modeling language doesn’t fit
• Close and continual interaction with external systems, such
as electrical and mechanical systems
– Examples: embedded systems, space ground systems
– Result: models need to capture interfaces to external systems
• Strict operating requirements
– Examples: real-time data processing, robust fault-tolerance
– Result: models need to be analyzable, executable
Model-Driven Engineering
• Model-driven engineering (MDE) combines domain-specific
languages (DSLs) with model interpreters
– Metamodels define elements, relationships, views, and constraints
– Model interpreters leverage domain-specific models for analysis,
generation, and transformation
8
Challenge: Constructing Interpreters
1. Designing, developing, and
maintaining DSLs and interpreters is
difficult and expensive
– A model interpreter must be
constructed for each analysis that
will be applied to a model
– Reusing model interpreters for
different DSLs is hard
– Little guidance exists on how to
construct DSLs and interpreters
– The semantics applied to models
are opaque (embedded in code)
– Requires particular types of
expertise
– Common topic of research papers
in the modeling community
Model Interpreter Implementation Tasks
1. Find a computational theory that derives the
relevant properties
2. Discover the semantic relationships between
the constructs present in the architectural
models and those present in the analysis
models
3. Determine the compatibility between the
assumptions and constraints of the architectural
models and the analysis models, and resolve
conflicts
4. Implement a model interpreter that executes a
sequence of operations to transform an
architectural model into an analysis model
5. Verify the correctness of the transformation
9
Simplifying Insight
Automatically synthesize
domain-specific model
interpreters the same way
that domain-specific model
editors are synthesized
Solution: Model Interpreter Frameworks
• Use a model interpreter framework to implement
domain-specific analysis
– Implements a mapping from metamodel types to target
platform types
– Configured via plug-ins generated from a metamodel
The eXtensible Toolsuite
for Evaluation of
Architectural Models
• A MDE platform for software architectures
• Includes:
– A specialized metamodeling language and metamodel editor
– Two metamodel interpreters with paired model interpreter
frameworks (model editor and simulation generator)
– Example metamodels and framework extensions
• Provides the extensibility to easily accommodate both new
modeling language features and new architectural analyses
XTEAM Usage
• Providing design rationale
• Weighing architectural
trade-offs
• Discovering emergent
behavior of component
assemblies
• Generating test cases and
validating component
implementations
XTEAM Simulation Generator
XTEAM Benefits
• Reduced implementation
effort
– Effort saved through code
generation and reuse
– Quantified by:
• Lines of generated
interpreter code
• Total lines of reused
interpreter code
• Lines of generated code per
domain-specific type
• Lines of reused code per
domain-specific type
• Reduced maintenance
effort
– Due to relative ease of
performing DSL
modifications within a
metamodel rather than
within model interpreter
source code
– Quantified by number of
metamodel objects
altered vs. number of
classes, methods, and
SLOC altered
Implementation Effort Metrics
AADL
C2
Client/server
Ecore
Prism-MW
Myx.fw
Pipe-and-filter
Publish/subscribe
xADL Core
Total SLOC Generated
3000
2500
2000
1500
1000
500
15
20
25
30
35
40
SLOC Generated per Domain-Specific Type
COCOMO Estimates:
Nominal settings  23.4 person-months
Favorable settings  4.2 person-months
Maintenance Effort Metrics
30
25
20
15
10
Types Affected
Classes Altered
Methods Altered
5
0
Code Altered/ Type
Summary
• Building model interpreters to analyze and
generate code from domain-specific models is
hard
• Our methodology leverages a metamodel and
extensible interpreter frameworks to
automatically synthesize domain-specific model
analysis tools and code generators
For More Information
Visit the XTEAM website:
http://www-scf.usc.edu/~gedwards/xteam.html
XTEAM Publications:
George Edwards and Nenad Medvidovic, A Methodology and Framework for Creating
Domain-Specific Development Infrastructures, Proceedings of the 23rd IEEE ACM
International Conference on Automated Software Engineering (ASE), September 2008.
George Edwards, Chiyoung Seo, and Nenad Medvidovic, Model Interpreter Frameworks: A
Foundation for the Analysis of Domain-Specific Software Architectures, Journal of
Universal Computer Science (JUCS), Special Issue on Software Components,
Architectures and Reuse, 2008.
George Edwards, Chiyoung Seo, and Nenad Medvidovic, Construction of Analytic
Frameworks for Component-Based Architectures, Proceedings of the Brazilian
Symposium on Software Components, Architectures and Reuse (SBCARS), August 2007.
George Edwards, Sam Malek, and Nenad Medvidovic, Scenario-Driven Dynamic Analysis of
Distributed Architectures, Proceedings of the 10th International Conference on
Fundamental Approaches to Software Engineering (FASE), March 2007.
Thanks for your attention