Slide 1

McGill University School of Computer Science Ph.D. Candidate in the Modelling, Simulation and Design Lab MPM09 Explicit Transformation Modelling Thomas Khne, Gergely Mezei, Eugene Syriani, Hans Vangheluwe, Manuel Wimmer Towards Systematic Transformation Development

Slide 2

MPM09 OVERVIEW Motivation Running Example FSA semantics from PN simulation Explicit Transformation Modelling RAM process Higher-Order Transformation Adding animation Conclusion 2

Slide 3

MPM09 MOTIVATION Increase modellers productivity Raise level of abstraction of specification/design Lower impedance mismatch between modelling language and application domain Explicitely model transformations as 1 st class entities Easier modification of control flow + mapping + pattern matching at syntax and semantics levels Domain-specific transformation models //vs// generic pattern matching language Facilitate transformation evolution: higher-order transformation (HOT) 3

Slide 4

MPM09 RUNNING EXAMPLE Finite State Automata (FSA) Simulator from Petri-Net Simulator 4

Slide 5

MPM09 FSA & PETRI-NET META-MODELS Classes Multiplicities Attribute types Constraints 5

Slide 6

MPM09 FINITE STATE AUTOMATA TO PETRI-NET 6

Slide 7

MPM09 PETRI-NET OPERATIONAL SEMANTICS 7

Slide 8

MPM09 EXPLICIT TRANSFORMATION MODELLING Generic //vs// Customized pattern language Syntax: pattern specification adapted to languages invloved Design: exclude patterns that would not match Custom Meta-model //vs// Conformance check Have a separate meta-model for the pattern specification Have customized conformance checks Meta-model 8

Slide 9

MPM09 RAM PROCESS (quasi-)Automatically generated environment for pattern language 9 Input Meta-ModelOutput Meta-Model Relax Augment Modify Customized Pattern Meta-Model

Slide 10

MPM09 RAM PROCESS Relaxation Concretize abstract entities Reduce minimal multiplicity constraint Constraints (automatic?) Free form: no constraints Valid elements: ensure typing Valid multiplicities: enforce (relaxed) multiplicities Valid constraints: enforce (a subset of) meta-model constraints (quasi-)Automatically generated environment for pattern language 10

Slide 11

MPM09 RAM PROCESS Augmentation Type all entities to MT(pre/post)_Element Add MT specific attributes Labels Pivot passing Allow subtype matching * Other (e.g., isProcessed) Generic elements 11 (quasi-)Automatically generated environment for pattern language

Slide 12

MPM09 RAM PROCESS Modification Pre-condition pattern Attributes are of type constraint Post-condition pattern Attributes are of type action Modify concrete syntax Abstract classes Association ends Other (e.g., replace topological visual syntax constraints) 12 (quasi-)Automatically generated environment for pattern language

Slide 13

MPM09 RAM PROCESS Relax 13 (quasi-)Automatically generated environment for pattern language AugmentModify

Slide 14

MPM09 RAM PROCESS Customized pattern language meta-model 14

Slide 15

MPM09 RAM PROCESS Transformation Meta-Model 15

Slide 16

MPM09 HIGHER-ORDER TRANSFORMATION (HOT) Language evolution Adapting the transformation of an evolving language Transformation optimization/refactoring Modify for more efficient results Translational semantics From operational semantics, for more advance analysis oportunities Define meaning of transformation by mapping onto a standard one Separation of transformation concerns Multi-stage transformation is a special case 16

Slide 17

MPM09 ANIMATION SUPPORT 17

Slide 18

MPM09 ANIMATION SUPPORT !User modifies meta-model: correspondence links ?Evolve the transformation without changing current transformation Transform the transformation to include animation support 18

Slide 19

MPM09 ANIMATION SUPPORT Add animation rules in simulator 19

Slide 20

MPM09 ANIMATION SUPPORT Control flow level HOT rule 20

Slide 21

MPM09 ANIMATION SUPPORT Rule level HOT rule 21

Slide 22

MPM09 CONCLUSION 22 Proposed a systematic procedure for the development of model transformations Allows to: Offer customized specification of model transfromation rules Cleanly design higher-order transformation rules

Slide 23

MPM09 CONCLUSION 23 Side-effect of the contribution: Given 1.Semantic translation from MM 1 to MM 2 (model transformation) 2.Operational semantics of MM 2 (model transformation) 3.Alterate transformation 1 to support feed-back (HOT) ? Translation of FSA to PN PN simulator Alteration of the control flow of the PN simulator Alteration of the rules of the FSA to PN translation You get an operational semantics for MM 1