Model Execution and Meta Layout

Model Execution and Meta Layout in EclipseSnapshots of the KIELER Project

Christian Motika and Miro Sponemann

Real-Time Systems and Embedded Systems GroupDepartment of Computer Science

Christian-Albrechts-Universitat zu Kiel, Germany

Eclipse Summit, 08/27/2009

Eclipse Summit 2009 Model Execution and Meta Layout in Eclipse 1 / 33

IntroductionKIELER Execution Manager

Use Case: PtolemyUse Case: Model Railway

Part I: Model Execution in Eclipse

Eclipse Summit 2009 Model Execution in Eclipse 2 / 33



MotivationOverview

Motivation

I EMF and GMF are great frameworks for modeling in Eclipse

I Model implementations, model and diagram editors, ...

I Modeling:

1. Structural modelsI E.g., class diagrams, component diagrams, ...

2. Behavioral modelsI E.g., flow charts, state machines, data flow models, ...




MotivationOverview

Motivation (cont’d)

System models are virtual as opposed to physical systems

I System models

I Generate code or documentationI System analysis and verificationI Simulation runsI ...

I Idea: Flexible definition of semantics and swapping out ofsimulation computation

I Solution proposed: KIELER Execution Manager




MotivationOverview

Overview

I KIELER Execution Manager

I Use case: Ptolemy

I M2M transformationI Simulation engine

I Use case: Model Railway

I InstallationI Railway Controller DSLI SimpleRailCtrl editor (DEMO)

I Summary




DataComponentsGUI And ThreadsInterface

KIEM Components





KIEM GUI and Threads

I DataObserver vs.DataProducer

I Scheduling & DataPool

I Properties

I Execution Buttons

I Master





Simple Interface

1 public interface IDataComponent {23 public void initialize() throws KiemInitializationException;45 public void wrapup() throws KiemInitializationException;67 public boolean isProducer();89 public boolean isObserver();

1011 public JSONObject step(JSONObject jSONObject)12 throws KiemExecutionException;1314 }





Flexible Extensions

1 public abstract class DataComponent implements IDataComponent,2 IExecutableExtension {34 public boolean isMultiInstantiable() {return false;}56 public String[] provideFilterKeys() {return null;}78 public KiemProperty[] provideProperties() {return null;}9

10 public void checkProperties(KiemProperty[] properties)11 throws KiemPropertyException {}1213 public String[] provideInterfaceKeys() {return null;}1415 public boolean isHistoryObserver() {return false;}1617 public boolean isDeltaObserver() {return false;}1819 public boolean isMaster() {return false;}2021 }




M2M TransformationSimulation Engine

Overview


I Use case: Ptolemy




I Summary





What is Ptolemy?

I”The Ptolemy project studies heterogeneous modeling,simulation, and design of concurrent systems.“

Introduction to Ptolemy II, UC Berkeley

I Executable models to describe behavior of reactive systems

I Set of components interacting under a model of computation

I → Actor-Oriented Design





Ptolemy EMF Model

I Ptolemy models can be executable

I DTD of the PtolemyXML representation (MOML)

I Acquire EMF modelI M2M transformationI Execute Ptolemy modelsI Back mapping of data/states





Simulation with Ptolemy




InstallationRailway Controller DSLSimpleRailCtrl Editor

Overview


I Use case: Ptolemy




I Summary





Installation

I Standard model railway equipment combined withI Over 200 sensors and actuatorsI Controlled by distributed computer system





Track Layout





Train Movement

I Several track segments individually controlled

I Computers get sensor information (instantaneous trainpositions) and control voltage

I ⇒ Actions: SetSpeed, SetPointI ⇒ Trigger: EventContact, EventTimeout





EMF Meta Model





Generated GMF Editor





Eclipse Model Execution Demo

LIVE DEMO




SummaryTo Go Further

Summary

I DSLs in Eclipse represented by EMF models

I Often only implicit execution semantics!

I Ptolemy models can be executable

I Xtend M2M transformation helps making semantics explicit

I Other simulator DataComponents imaginable

I KIELER Execution Manager seamlessly integrates executioninto the Eclipse RCP




SummaryTo Go Further

To Go Further

UC BERKELEY, EECS DEPT.Ptolemy webpage.http://ptolemy.eecs.berkeley.edu/.

UNI KIEL, REAL-TIME AND EMBEDDED SYSTEMS GROUP.KIELER webpage.http://www.informatik.uni-kiel.de/en/rtsys/kieler/.

[2] [1]


IntroductionThe Interface

The Background

Part II: Meta Layout ofGraphical Models

Eclipse Summit 2009 Meta Layout of Graphical Models 23 / 33


The Background

Automatic Layout in GMFKIELER Meta Layout

Automatic Layout in GMFI GMF supports automatic layout...

I ...but is not very flexibleI No selection of different layout algorithmsI No customization of layout optionsI No deep layout of compound structures



The Background

Automatic Layout in GMFKIELER Meta Layout

KIELER Meta Layout

I Meta Layout: Allow fully flexible automatic diagram layout

I Contribute new layout algorithms using extension pointsI Customize layout options in the properties viewI Layout compound structures recursivelyI Layout different parts of a diagram with different options, or

even with different layout algorithms

I Development of special layout algorithms, e.g. for data flowdiagrams



The Background

Layout ProvidersCustom Layout OptionsCompound StructuresMeta Layout

Layout ProvidersExtension points are used to

I define diagram types

I state machine, class diagram, etc.

I assign diagram types and layout options to specific parts of aGMF diagram

I e.g. assign the “class diagram” type to the diagram edit partof a class diagram editor

I contribute new layout algorithms

I call these contributions layout providers

I define which diagram types are supported by the layout provider

I define new layout options and specify which options areunderstood by a layout provider



The Background


Custom Layout Options

I To be implemented soon...



The Background


Compound StructuresI Apply layout providers recursively



The Background


Meta Layout

Live Demo...



The Background

Diagram LayoutData Flow Diagrams

Diagram Layout

I Layout providers work on an internal graph structure generatedwith EMF

I Need to map the contents of a diagram to the internal structureand back

I Done by Layout Managers using the GEF command / requestpattern

I Currently only a layout manager for GMF is implemented

I Analyzes the edit part structure at runtimeI Recursively go into the contents of an edit part to explore

compound structures

I Extension to other diagram editor generation frameworks ispossible



The Background


Data Flow Diagrams

I Operators exchange data through ports

I Layout algorithms must respect these ports when routingconnections



The Background


Data Flow Diagrams: Special Algorithm

I Developed a special algorithm to layout data flow diagrams

I Internal graph structure supports ports

I Integrated in the KIELER Meta Layout

I Also available as stand-alone library, successfully applied toPtolemy



The BackgroundSummary

Summary

I KIELER Meta Layout provides flexible automatic diagram layout

I Customize layout algorithms and layout options

I Current implementation is able to layout all GMF diagrams

I Implemented a special layout algorithm for data flow diagrams


Model Execution and Meta Layout

Education

Transcript of Model Execution and Meta Layout