XTEAM: Automated Synthesis of Domain-Specific Code Generators George Edwards [email protected] Introduction to the Center for Systems and Software Engineering.

XTEAM: Automated Synthesis ofDomain-Specific Code

Generators

George [email protected]

Introduction to the Center for Systems and Software

Engineering

Brief Introduction to CSSE• Established in 1993 as CSE

– Merged in 2006 with SAE to form CSSE

• Focus on research and teaching– MSCS with emphasis in SE– Graduate Certificate in SAE– Two Certificate and Advanced Degree programs

• Objective is to solve scientific problems with industrial relevance

CSSE Affiliates• 22 industrial affiliates

– 8 from aerospace industry– 12 from other industries– 2 international

• 12 government affiliates– 6 government agencies

• incl. US Army TACOM and Research Labs

– 6 FFRDCs and consortia

• 7 academic affiliates– 5 universities– 2 research centers

• 3 visiting associates

Research Focus Areas

• Architecture• Economics• Process• Testing & analysis

Software Architecture Research

• Modeling• Analysis• Implementation• Deployment• Dynamic adaptation• Architectural styles• Software connectors• Architectural recovery

XTEAM Project Overview

• Processes, notations, tools, and designs that enable automated synthesis of end-to-end domain-specific toolsets for software architecture modeling, analysis, and code generation

• Example application areas:– Systems-of-systems and ultra-large-scale

systems– Safety- and mission-critical systems– Cloud and grid systems

Domain-Specific Languages (DSLs)

• Customized for a particular family of problems(the domain)

• Concise and intuitive– No missing or extra

features– Capture patterns– Enforce constraints– Use native symbols and

terms

• Can be modified, evolved, and composed

Model-Driven Engineering (MDE)• MDE leverages DSLs for architecture modeling

– Metamodels define DSL syntax (types, properties, views, and constraints)– Model interpreters define DSL semantics (analysis, code generation,

etc.)

• MDE platforms provide (some) tool support– Metamodel editor with built-in metamodeling language– Metamodel interpreter that configures a metaprogrammable model

editor

Metamodel Editor

Metamodel Editor

Metaprogrammable Model Editor

Metaprogrammable Model Editor

Model Execution

Environment (Semantic Domain)

MetamodelInterpreter

ModelInterpreter

Off-the-shelf

Auto generated

Built by engineer

Domain-Specific Model

Domain-Specific Model

Domain-Specific ModelMetamodel

Metamodeling Language

Domain-Specific Language

Executable Model

Executable Model

Executable Model

8

“The difficulty of building and maintaining a DSM solutionstems essentially from the complexity of the mapping between the concept instances expressed in the DSML and the code that has to be generated.”

A. L. Santos et al. Automating the Construction of Domain-Specific Modeling Languages for Object-Oriented Frameworks. Journal of Software and Systems, 2010.

Problems with MDE• Building and

maintaining code generators for DSLs is inherently difficult– High design complexity– Disproportionate

maintenance and evolution costs

– Hard to verify correctness– Redundant development

effort– Opaque semantics

embedded in source code

“There is a fundamental problemin keeping the model interpreters up to date with metamodel changes ... Current practice requires each model interpreter to be modified manually after each metamodel schema change. This can be a very time-consuming and error prone task for complex model interpreters of considerable size.”

Jing Zhang. Metamodel-Driven Model Interpreter Evolution. Conference on Object Oriented Programming Systems Languages and Applications, 2005.

“Checking mathematicalproperties like correctness or completeness of transformations based on common programminglanguages is very difficult...”

I. Malavolta, H. Muccini, P. Pelliccione, and D. Tamburri. Providing Architectural Languages and Tools Interoperability through Model Transformation Technologies. IEEE Transactions on Software Engineering, 2009.

“The state-of-the-art of modelinterpreter writing needs to be advanced to enhance thereusability and maintainability of this software...”

G. Karsai. Structured Specification of Model Interpreters. Engineering of Computer-Based Systems, 1999.

“Writing translators by hand...in addition to being inefficient, has yet another serious drawback: the semantic mapping between the input and the output is vaguely specified...[Building model interpreters] is the most time consuming and error prone phase of the MIC approach...”

G. Karsai, A. Agrawal, F. Shi, J. Sprinkle. On the Use of Graph Transformation in the Formal Specification of Model Interpreters. Journal of Universal Computer Science, 2003.

FCS: A Real World Example• > $100 billion US Army

program

• DSL with hundreds of types– Modified on a daily basis

• Automated generation of:– Discrete event simulations– Middleware configuration files– Fault trees– Spreadsheets for documentation

Tool building and maintenance required approximately five full-time MDE experts

Synthesize domain-specific code generators using the same mechanisms that have proven successful for synthesizing domain-specific model editors.

XTEAM Solution Approach

Metamodel Editor

Metamodel Editor

Metamodel

Metamodel

Metamodel Interpreter AMetamodel

Interpreter A

Model Editor Framework

Model Editor Framework

Metamodel Interpreter BMetamodel

Interpreter B

Model Interpreter Framework

Model Interpreter Framework

Target PlatformTarget

Platform

Presentation Semantics

Platform Semantics

Application Model(Abstract

Representation)

Application Model(Abstract

Representation) Executable

ModelExecutable

Model

Types and Views

Presentation Properties

Platform Properties

Presentation Logic

Presentation Rules

Code Generation Logic

Code Generation Rules

Target DisplayTarget Display

Visualized Model

Visualized Model

XTEAM Toolset

12

Windows/C++/Visual Studio

Fedora Linux

SVN Repository

JVM

Prism-MWLightweight

Architectural Middleware

XDEVS Discrete Event

Simulation Engine

GMEModel Editor

GMEModel Editor

XTEAM Metalanguage

XDEVS Metainterpreter

Prism-MW Metainterpreter

GME Metainterpreter

XDEVS MIF

Prism-MW MIF

Prism-MW Applications

(Java)

Prism-MW MIF Extensions

C++

XDEVS MIF Extensions

XDEVS Simulations

(C++)

XTEAMModels

(XML/HTTP)

XTEAMMetamodels(XML/HTTP)

Off-the-Shelf Component

XTEAM Component

Connector

Languages and Platforms

XTEAM DSLs

XTEAM Simulations

• Providing design rationale

• Weighing architectural trade-offs

• Discovering emergent behavior of component assemblies

• Validating component implementations

13

Providing Design Rationale

• Architects rely on intuition and experience to make important decisions early in the design phase– What architectural style to

use– How to allocate functionality

among subsystems– What types of connectors to

use

• XTEAM allows architects to rationalize such decisions with experimental evidence– Model confidence level: Low

Peer-to-peerArchitecture

Client-Server Architecture

PotentialWorkload

0

500

1000

1500

2000

2500

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

Resp

on

se T

ime (m

s)

Invocation

Request File Response Time

Weighing Architectural Trade-offs• Nearly all architectural decisions come down to trade-offs

between multiple desirable properties– Emphasis on one system property may yield diminishing returns

• XTEAM allows architects to evaluate design alternatives in terms of their impact on multiple non-functional properties– Model confidence level: Medium

Replication of components

Decreases response time

Consumes more battery power

0

500

1000

1500

2000

2500

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

Resp

on

se T

ime (m

s)

Invocation

Request File Response Time

S…

22000

22500

23000

23500

24000

0 58944 104612 175249 239514 287914 352870 415963 480308 529223

Rem

ain

ing

En

erg

y (m

J)

Time (ms)

Host A (iPAQ) Battery Power

Evaluating Component Assemblies

• Contemporary large-scale distributed systems contain numerous off-the-shelf components– Detailed information about the behaviors and properties of

individual components may be known– Component assemblies may exhibit unforeseen behavior due to

subtle interactions between constituent components• XTEAM allows an architect to determine the emergent

properties of a composed system– Model confidence level: High

Off-the-shelf components Highly accurate parameterizationDetermination of

emergent properties

Incremental System Validation• Individual component

implementations may become available in a piecemeal fashion

• XTEAM allows architects to1. Immediately incorporate

component implementations into a simulated system, increasing the accuracy of analysis results

2. Rapidly test individual components in the context of a wide variety of operational scenarios

– Model confidence level: High

Componentimplementation

available

Invoke implementation from behavior model

Integrated simulation and test environment

For More Information

• CSSE Website:http://csse.usc.edu/csse/

• XTEAM Website:http://www-scf.usc.edu/~gedwards/xteam.html