Upload
ed-seidewitz
View
4.058
Download
14
Tags:
Embed Size (px)
DESCRIPTION
This is a one day workshop presentation, primarily on the new OMG Foundational UML specification for executable model semantics, but also discussing extensions for executable SysML (System Modeling Language) models.
Citation preview
February 2010
Copyright © 2010 Data Access Technologies, Inc.(Model Driven Solutions)
A division of Data Access Technologies, Inc.
Executable UML and SysML
Ed Seidewitz
Page 2 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Agenda
I. Introduction
II. Foundation for Executable UML
III. Extensions for Executable SysML
Appendix: Specifying Execution Semantics
Page 3 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
I. Introduction
Objectives
• To motivate the need for standardized model execution.
• To learn about the latest OMG standards related to model execution.
Page 4 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
I. Introduction
Topics
A. Why Model?
B. Why Execute Models?
C. Why Standardize?
D. A Motivating Example
Page 5 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
A. Why Model?
• A model is a set of statements in some modeling language made about some system or domain.– Standard modeling languages: Unified Modeling Language (UML),
Business Process Modeling Notation (BPMN), Systems Modeling Language (SysML), Service Oriented Architecture Modeling Language (SoaML), etc.
• A model may be used to describe a domain or system under study or to specify a (business, software and/or hardware) system to be built.– Descriptive models are generally used for analysis.– Specification models are generally used for engineering.
• Models are intended to represent and communicate the results of analyses and proposals for new syntheses.– No model can represent everything – but, to be useful, a model must
effectively promote general understanding and communicate important details.
Page 6 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
B. Why Execute Models?
• A model may specify the behavior of a system, that is, how the system interacts with external entities and changes its state over time.
• A behavioral model is executable if it is complete enough that the specified behavior can be enacted or simulated by an automated execution tool.
• Model execution may be used to:– Explore possible (desirable and undesirable) behaviors of a system– Validate the behavioral specification for a system– Actually act as the implementation of the system (particularly for
business processes or software systems)
Page 7 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Developers provide feedback to the architects(maybe)
Modeling for Software Development
But…• It is hard to validate the correctness of the models before development.• The developers may not follow the models, without providing feedback.• It is hard to keep the models and development artifacts in sync during
development (and maintenance).
How it usually works without executable modelsArchitects give models to developers
Developers create artifacts based on the models(maybe)
Architects create the models
Page 8 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Executable Modeling for Software Development
Architects validate the models by executing them in a simulated test environment
Technologists specify the implementation platform
The models are provisioned as executing artifacts on the target platform
Architects create the models
The models are the source code.
Using a standard-conforming UML
modeling tool
Using a standard-conforming UML
modeling tool
Using a standard-conforming UML
execution tool
Using a standard-conforming UML
execution tool
How it works with executable models
Page 9 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
• Hardware and software engineers develop components to satisfy the requirements.
• Test engineers develop the test environment to verify the requirements.
Executable Modeling for System Engineering
System engineers analyze, simulate and validate the system design, and allocate
requirements to components.
Using a standard-conforming SysML
modeling tool
Using a standard-conforming SysML
modeling tool
Using a standard-conforming SysML
execution tool
Using a standard-conforming SysML
execution tool
System engineers create the models
Execution artifacts could include:• System behavior• Timing• Statistics
Execution artifacts could include:• System behavior• Timing• Statistics
Models can include both hardware and
software components.
Models can include both hardware and
software components.
Page 10 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
C. Why Standardize?• Semantic interoperability and consistency across modeling and
execution tools requires standardization– UML model execution tools are available, but UML execution semantics have
not previously been precisely standardized.• Mentor Graphics BridgePoint• Kennedy-Carter xUML• Rational Rose Real Time and Rhapsody• Kabira Fluency
– System engineering model execution tools are available, but have not previously provided a complete, standardized modeling capability
• Vitech CORE Enhanced Functional Flow Block Diagrams (EEFBDs)
• OMG standards are now filling these gaps– Unified Modeling Language (UML)– Systems Modeling Language (SysML)– Executable UML Foundation (fUML)– UML Action Language (Alf)
Page 11 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Unified Modeling Language (UML)
The Unified Modeling Language (UML) is a graphical language for modeling the structure, behavior and interactions of software, hardware and business systems, standardized by the Object Management Group (OMG).
• UML Version 1.1 (first standard) – November 1997
• UML Version 2.0 – August 2005
• UML Version 2.3 (current standard) – September 2009 (Beta)
• UML Version 2.4 – May 2010 (planned)
• UML Version 2.5 (spec simplification) – December 2010 (planned)
Page 12 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Systems Modeling Language (SysML)
The Systems Modeling Language (SysML) is a general-purpose modeling language for systems engineering applications that supports the specification, analysis, design, and verification and validation of a broad range of complex systems. These systems may include hardware, software, information, processes, personnel, and facilities.
• SysML is a specialized usage or profile of UML.
• It is the result of a collaborative effort between OMG and the International Council on System Engineering (INCOSE).
• SysML Version 1.0 – September 2007
• SysML Version 1.1 – November 2008
• SysML Version 1.2 Beta 2 – August 2009
Page 13 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Executable UML Foundation (fUML)
Foundational UML (fUML) is an executable subset of standard UML that can be used to define, in an operational style, the structural and behavioral semantics of systems.
• OMG RFP for the Semantics of a Foundational Subset for Executable UML Models – Issued April 2005
• fUML Version 1.0 Beta 1 – November 2008
• fUML Version 1.0 Beta 2 – September 2009
• fUML Version 1.0 Beta 3 (finalized) – February 2010
Page 14 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
UML Action Language (Alf)
The Action Language for Foundational UML (Alf) is a textual surface representation for UML modeling elements with the primary of acting as the surface notation for specifying executable (fUML) behaviors within an overall graphical UML model.
• OMG RFP for Concrete Syntax for a UML Action Language – Issued September 2008
• Two initial submissions – August 2009
• Joint revised submission – February 2010
• Second revised submission – May 2010 (planned)
Page 15 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
D. A Motivating Example
• Simplification of the Hybrid SUV Sample Problem from Annex B of the SysML Specification.
• Example behavioral models (using only standard UML without SysML extensions at this point)– State machines– Activities
• The goal: automated execution and analysis
Page 16 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Hybrid SUV Operational States
A state machine abstracts system behavior into a finite number of states.
A state machine abstracts system behavior into a finite number of states.
The system is modeled as having discrete transitions between the states.
The system is modeled as having discrete transitions between the states.
A transition may trigger further system behavior…A transition may trigger further system behavior…
…or system behavior may be dependent on the current state/
…or system behavior may be dependent on the current state/
Page 17 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
“Accelerate” Activity
An activity specifies behavior as the coordinated execution of a set of subordinate actions.
An activity specifies behavior as the coordinated execution of a set of subordinate actions.
An action in one activity may call another activity.
An action in one activity may call another activity.
Data and control flow between the various actions.
Data and control flow between the various actions.
Page 18 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
“Provide Power” Activity
Other actions provide various data and computational functions.
Other actions provide various data and computational functions.
Page 19 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
“Proportion Power” Activity
Full executability requires complete specification of all behavior and computation.
Full executability requires complete specification of all behavior and computation.
However, after a certain level of detail, it is much more convenient to use a textual rather than graphical notation (whether mathematical constraints or procedural action language.)
However, after a certain level of detail, it is much more convenient to use a textual rather than graphical notation (whether mathematical constraints or procedural action language.)
Page 20 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
The Goal: Automated Analysis
Page 21 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
II. Foundation for Executable UML
Objectives
• To understand the intent of the fUML standard.
• To provide an overview of fUML as the foundational language for executable UML models.
• To understand how models are given executable semantics.
Page 22 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
II. Foundation for Executable UML
Topics
A. Executable UML Foundation Standard
B. The fUML Subset
C. Foundation Model Library
Page 23 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
A. Executable UML Foundation Standard
• Submitters and Supporters
• Objectives
• Key Components
• Semantics
• Conformance
Semantics of a Foundational Subset for Executable UML Models
Page 24 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
• Submitters
– CARE Technologies
– International Business Machines
– Kennedy Carter
– Lockheed Martin
– Mentor Graphics
– Model Driven Solutions
• Supporters
– U.S. National Institute of Standards and Technology
– 88Solutions Corporation
– CEA LIST/LISE
– NASA Jet Propulsion Laboratory
Submitters and Supporters (Foundation)
Page 25 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Submitters and Supporters (Action Language)
• Submitters
– Model Driven Solutions
– Mentor Graphics
– 88solutions Corporation
– No Magic
– International Business Machines
– Visumpoint
• Supporters
– Ericsson AB
Page 26 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Objectives
1. To enable a chain of tools that support the construction, verification, translation, and execution of computationally complete executable models based on a foundational subset of UML.
2. To provide a general and standardized facility for specifying the structural and behavioral semantics of MOF-based modeling languages. Also, the same mechanisms can be used to define the semantics of domain-specific stereotypes defined in profiles as well as the concepts of languages other than UML.
Page 27 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Key Components
• Foundational UML Subset (fUML) – A computationally complete subset of the abstract syntax of UML (Version 2.3)– Kernel – Basic object-oriented capabilities
– Common Behavior – General behavior and asynchronous communication
– Activities – Activity modeling, including structured activities (but not including variables, exceptions, swimlanes, streaming or other “higher level” activity modeling)
• Execution Model – A model of the execution semantics of user models within the fUML subset
• Foundational Model Library– Primitive Types – Boolean, String, Integer, Unlimited Natural
– Primitive Behaviors – Boolean, String and Arithmetic Functions
– Basic Input/Output – Based on the concept of “Channels”
Page 28 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Composite Structure
Semantics
Composite Structure
Semantics
Complete Activity Model
Semantics
Complete Activity Model
Semantics
State Machine Semantics
State Machine Semantics
Semantics
Non-Executable
Model Semantics
Non-Executable
Model Semantics
The semantics of fUML provide the foundation for formally specifying the (execution) semantics of the rest of UML.
Some areas of UML (e.g., use case and requirements models) may not be best formalized
based on an executable semantics foundation.
Some areas of UML (e.g., use case and requirements models) may not be best formalized
based on an executable semantics foundation.
Interaction Model
Semantics
Interaction Model
Semantics
Foundational SemanticsFoundational Semantics
fUML operational semantics are specified as an execution model written in fUML itself.
fUML operational semantics are specified as an execution model written in fUML itself.
Base SemanticsBase Semantics
The base semantics of the subset of fUML used in the execution
model are specified using formal logic.
The base semantics of the subset of fUML used in the execution
model are specified using formal logic.
Page 29 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Conformance
• Fundamental aspects1. Syntactic Conformance. A conforming model must be restricted to
the abstract syntax subset defined for fUML.2. Semantic Conformance. A conforming execution tool must provide
execution semantics for a conforming model consistent with the semantics specified for fUML.
• Levels (parallel for syntax and semantics)L1. Kernel, BasicBehaviors, Communications, Loci (semantics only)L2. IntermediateActivities, BasicActions, IntermediateActionsL3. CompleteStructuredActivities, ExtraStructuredActivities,
CompleteActions
Page 30 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
B. The fUML Subset
i. Activities
ii. Actions
iii. Structure
iv. Asynchronous Communication
Page 31 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
i. Activities
• Activities and Parameters
• Actions and Flows
• Textual Notation
• Tokens
• Offers
• Control Nodes
• Structured Nodes
Page 32 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Activities and Parameters
An activity is a specification of behavior as the coordinated execution of subordinate actions, using a control and data flow model.
An activity is a specification of behavior as the coordinated execution of subordinate actions, using a control and data flow model.
An activity may have input, output and return parameters.An activity may have input, output and return parameters.
The parameters have corresponding activity parameter node on the boundary of the diagrammatic representation of an activity.
The parameters have corresponding activity parameter node on the boundary of the diagrammatic representation of an activity.
Page 33 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Actions and Flows
An action is a fundamental unit of executable behavior within an activity.
An action is a fundamental unit of executable behavior within an activity.
A pin is an activity node that either accepts input to or provides output from an action.
A pin is an activity node that either accepts input to or provides output from an action.
An object flow provides a path for passing objects or data.
An object flow provides a path for passing objects or data.
A control flow specifies the sequencing of actions.
A control flow specifies the sequencing of actions.
An activity diagram is a graph structure consisting of activity nodes connected by activity edges.
An activity diagram is a graph structure consisting of activity nodes connected by activity edges.
Page 34 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Textual Notation
activity DoSomething(in input: Integer, out output Integer): Integer { output = A(input); return B();}
activity DoSomething(in input: Integer, out output Integer): Integer { output = A(input); return B();}
Alf behavioral notation maps to fUML activity models.
Alf behavioral notation maps to fUML activity models.
The semantics of the Alf notation is defined by its mapping to fUML
The semantics of the Alf notation is defined by its mapping to fUML
Page 35 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Tokens
a = DoSomething(1, b);a = DoSomething(1, b);
A token is a container for an object, datum or locus of control that may be present at an activity node.
A token is a container for an object, datum or locus of control that may be present at an activity node.
The activity is invoked with an argument of 1 for its input parameter.
The activity is invoked with an argument of 1 for its input parameter.
An object token with a value of 1 is placed on the input activity parameter node.
An object token with a value of 1 is placed on the input activity parameter node.
The object token flows to the input pin of action A along the object flow.
The object token flows to the input pin of action A along the object flow.
Action A fires and produces an object token on its output pin.
Action A fires and produces an object token on its output pin.
The object token flows to the output activity parameter node along the object flow.
The object token flows to the output activity parameter node along the object flow.
When it is done, action A produces a control token, which flows to action B along the control flow.
When it is done, action A produces a control token, which flows to action B along the control flow.
Action B accepts the control token and fires, producing an object token on its output pin.
Action B accepts the control token and fires, producing an object token on its output pin.
The object token flows to the output activity parameter node along the object flow.
The object token flows to the output activity parameter node along the object flow.
Values on the output activity parameter nodes are copied to the output arguments.
Values on the output activity parameter nodes are copied to the output arguments.
Page 36 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Offers
An output pin offers its tokens to the targets of all outgoing object flows.
An output pin offers its tokens to the targets of all outgoing object flows.
A single token can only flow to one target. If two competing targets are both ready to accept an offer for the same token, it is indeterminate which will get the token.
A single token can only flow to one target. If two competing targets are both ready to accept an offer for the same token, it is indeterminate which will get the token.
Note: fUML semantics do not guarantee “liveliness” or “fairness” in the execution of actions competing for tokens.
Note: fUML semantics do not guarantee “liveliness” or “fairness” in the execution of actions competing for tokens.
Actions with no control constraints execute concurrently. This means that they may execute in parallel – or they may execute sequentially in any order.
Actions with no control constraints execute concurrently. This means that they may execute in parallel – or they may execute sequentially in any order.
Page 37 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Fork and Join Nodes
order = 'Create Order'();@parallel { 'Fulfill Order'(order); 'Invoice Order'(order);}'Close Out Order'(order);
order = 'Create Order'();@parallel { 'Fulfill Order'(order); 'Invoice Order'(order);}'Close Out Order'(order);
A fork node copies the tokens it is offered, and offers a copy on each outgoing flow.
A fork node copies the tokens it is offered, and offers a copy on each outgoing flow.
A join node waits for a token to be offered on all incoming flows and then offers tokens on its outgoing flow.
A join node waits for a token to be offered on all incoming flows and then offers tokens on its outgoing flow.
In the Alf textual notation, forks and joins are implicit in the parallel block notation.
In the Alf textual notation, forks and joins are implicit in the parallel block notation.
Note: Alf does not actually provide any notation for competition for tokens on output pins. A fork node is always inserted for multiple flows out of any output pin.
Note: Alf does not actually provide any notation for competition for tokens on output pins. A fork node is always inserted for multiple flows out of any output pin.
Page 38 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Control NodesAn initial node generates a single control token.
An initial node generates a single control token.A merge node
passes on any tokens it receives.
A merge node passes on any tokens it receives.
A decision node routes tokens based on a decision input value.
A decision node routes tokens based on a decision input value. An activity final
node terminates the activity when it receives a token.
An activity final node terminates the activity when it receives a token.
A control node is an activity node used to coordinate the flow of (the offers for) tokens between other nodes.
A control node is an activity node used to coordinate the flow of (the offers for) tokens between other nodes.
Note: This construction is necessary so a “card” token is available for each iteration.
Note: This construction is necessary so a “card” token is available for each iteration.
Note: Since the decision node “gates” control flow, it must be provided with an incoming control token.
Note: Since the decision node “gates” control flow, it must be provided with an incoming control token.
Page 39 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Structured Nodes
card = 'Select Credit Card'();
do { charge = 'Create Credit Card Charge'(card);
if ('Check Charge Approval'(charge)) { declined = false; 'Notify Customer of Approval'(charge); } else { declined = true; 'Notify Customer of Denial'(charge); }
} while (declined);
card = 'Select Credit Card'();
do { charge = 'Create Credit Card Charge'(card);
if ('Check Charge Approval'(charge)) { declined = false; 'Notify Customer of Approval'(charge); } else { declined = true; 'Notify Customer of Denial'(charge); }
} while (declined);
A structured node is an activity node used to group subordinate nodes into a control structure.
A structured node is an activity node used to group subordinate nodes into a control structure.
An loop node iterates the execution of its body while a condition is true.
An loop node iterates the execution of its body while a condition is true.
By default, the condition is tested after execution of the body.
By default, the condition is tested after execution of the body.
A conditional node executes one clause or another based on the result of a test (or tests).
A conditional node executes one clause or another based on the result of a test (or tests).
Inputs to the loop node initialize loop variables available across all iterations of the loop.
Inputs to the loop node initialize loop variables available across all iterations of the loop.
Note: There is no normative UML graphical notation for loop or conditional nodes.
Note: There is no normative UML graphical notation for loop or conditional nodes.
Page 40 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Expansion Regions
'Get Outstanding Orders'(customer) -> select order ('Is Delinquent?'(order)) -> iterate order ('Refer for Collection'(order));
'Get Outstanding Orders'(customer) -> select order ('Is Delinquent?'(order)) -> iterate order ('Refer for Collection'(order));
An expansion region is used to apply subordinate actions on all members of an input collection
An expansion region is used to apply subordinate actions on all members of an input collection
A parallel expansion region applies nested behavior concurrently to all collection elements.
A parallel expansion region applies nested behavior concurrently to all collection elements.
An iterative expansion region applies nested behavior sequentially to all collection elements.
An iterative expansion region applies nested behavior sequentially to all collection elements.
Alf provides specialized notation that maps to typical uses of expansion regions.
Alf provides specialized notation that maps to typical uses of expansion regions.
Page 41 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
ii. Actions
• Invocation Actions
• Object Actions
• Structural Feature Actions
• Link Actions
NOTE: Some of these actions will be discussed in more detail later.NOTE: Some of these actions will be discussed in more detail later.
Page 42 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Invocation Actions
• Call Behavior
– Calling an activity
– Calling a primitive behavior
• Call Operation
• Send Signal
• Accept Event
PlaceOrder(customer, product)
Max(throttle, limit) count + quantity
order.addProduct(product, quantity)
vehicle.EngageBrake(pressure)
accept (signal: EngageBrake)
Page 43 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Object Actions
• Value Specification
• Create Object
• Destroy Object
• Test Identity
• Read Self
• Read Extent
• Read Is Classified Object
• Reclassify Object
1 true "Hello"
new Order()
order.destroy()
order == myOrder name != customerName
this
Order.allInstances()
vehicle instanceof Car car hastype Hatchback
reclassify order from PendingOrder to ClosedOrder
Page 44 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Structural Feature Actions
• Read Structural Feature
• Add Structural Feature Value
• Remove Structural Feature Value
• Clear Structural Feature Value
order.customer
order.lineItems->add(item)
order.lineItems->remove(item)
order.card = null
Page 45 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Link Actions
• Read Link
• Create Link
• Destroy Link
• Clear Association
Owns->select person (house=>thisHouse)
Owns->add(person=>jack, house=>newHouse)
Owns->remove(person=>jack,
house=>oldHouse)
Owns->clear(jack)
Page 46 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
iii. Structure
• Structural and Behavioral Models
• Classes
• Associations
Page 47 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Structural and Behavioral Models
• A structural model (e.g., a class model) specifies the relevant instances in a domain that may exist at any one point in time.– Structural semantics define how a structural model constrains
allowable instances.
• A behavioral model (e.g., an activity model) specifies behavior over time– Behavioral semantics define how a behavioral model changes the
state of instances over time.
Page 48 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes
• Attributes
• Data Types
• Primitive Types
• Operations and Methods
• Structural Semantics
• Behavioral Semantics
A class is a classifier of objects that persist in the extent of the class, with an identity that is independent of the value of their attributes at any one time.
A class is a classifier of objects that persist in the extent of the class, with an identity that is independent of the value of their attributes at any one time.
Page 49 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes and Attributes
A class may have attributes whose types are primitive, data types or other classes.
A class may have attributes whose types are primitive, data types or other classes.
A referential attribute, whose type is a class, is conventionally notated as an association with a class-owned association end.
A referential attribute, whose type is a class, is conventionally notated as an association with a class-owned association end.
A bidirectional association results in corresponding referential attributes on both associated classes.
A bidirectional association results in corresponding referential attributes on both associated classes.
Page 50 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Data Types
A data type is a classifier of transient data values whose identity is based on the values of their attributes.
A data type is a classifier of transient data values whose identity is based on the values of their attributes.
Data types may have attributes, but not operations.
Data types may have attributes, but not operations.
Page 51 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Primitive Types
• From UML 2 Auxiliary Constructs– Boolean– Integer– UnlimitedNatural– String
• To be added for Alf– Bit strings– Real/floating point– Fixed point (?)
Page 52 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes: Operations and Methods
An operation specifies a behavior that may be synchronously invoked on an instance of a class.
An operation specifies a behavior that may be synchronously invoked on an instance of a class.
A method defines that actual behavior that is invoked.
A method defines that actual behavior that is invoked.
Page 53 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes: Structural Semantics
Structural semantics specify how a structural model constrains allowable instances.
Structural semantics specify how a structural model constrains allowable instances.
Objects are instances of classes with values for each attribute.
Objects are instances of classes with values for each attribute.
Class-owned association ends are structural features with values, like attributes.
Class-owned association ends are structural features with values, like attributes.
fUML does not actually give semantics to an association with class-owned ends, only to the ends as structural features.
fUML does not actually give semantics to an association with class-owned ends, only to the ends as structural features.
Note: fUML does provide “reified” semantics for associations that own their own ends, as will be discussed later.
Note: fUML does provide “reified” semantics for associations that own their own ends, as will be discussed later.
Page 54 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes: Behavioral Semantics
• Creating an Order
• Adding a Line Item
• Canceling an Order
Behavioral semantics specify how a behavioral model changes the state of instances over time.Behavioral semantics specify how a behavioral model changes the state of instances over time.
Page 55 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Creating an Order
order = new Order (customer, today)
Before After
@createpublic Order (customer: Customer, in datePlaced: Date) { this.datePlaced = datePlaced; this.totalAmount = new Money(dollars=>0, cents=>0); this.customer = customer;
this.customer.orders->add(this);}
@createpublic Order (customer: Customer, in datePlaced: Date) { this.datePlaced = datePlaced; this.totalAmount = new Money(dollars=>0, cents=>0); this.customer = customer;
this.customer.orders->add(this);}
A new object is created and the constructor operation is invoked.
A new object is created and the constructor operation is invoked.
The constructor initializes the new order’s attribute values…
The constructor initializes the new order’s attribute values…
…and adds the order to the customer’s list.…and adds the order to the customer’s list.
Page 56 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Adding a Line Item
Before Afterorder.addProduct (product, 2)
public addProduct (in product: Product, in quantity: Integer) { lineItem = new LineItem(product, quantity); this.lineItems->add(lineItem); this.totalAmount = MoneyFunctions::Add (this.totalAmount, lineItem.amount);}
public addProduct (in product: Product, in quantity: Integer) { lineItem = new LineItem(product, quantity); this.lineItems->add(lineItem); this.totalAmount = MoneyFunctions::Add (this.totalAmount, lineItem.amount);}
The method for the operation creates a new line item object…
The method for the operation creates a new line item object…
The addProduct operation is invoked on an existing Order object.
The addProduct operation is invoked on an existing Order object.
…adds the new object to the list of line items for the order…
…adds the new object to the list of line items for the order…
…and updates the total order amount.
…and updates the total order amount.
Page 57 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Canceling an Order
Before Afterorder.cancel()
@destroypublic cancel() { this.customer.orders->remove(this);}
@destroypublic cancel() { this.customer.orders->remove(this);}
A destructor operation is invoked, after which the order object is destroyed.
A destructor operation is invoked, after which the order object is destroyed.
Because line items are aggregated by composition, they are destroyed, too.
Because line items are aggregated by composition, they are destroyed, too.
References to the destroyed object must be explicitly removed.
References to the destroyed object must be explicitly removed.
Page 58 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Associations
• Classes and Associations
• Structural Semantics
• Behavioral Semantics
An association is a classifier whose instances are links that relate other instances.An association is a classifier whose instances are links that relate other instances.
Page 59 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classes and Associations
An association (that owns its ends) is a classifier of persistent links between the associated classes, which exist in the extent of the association.
An association (that owns its ends) is a classifier of persistent links between the associated classes, which exist in the extent of the association.
Page 60 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Associations: Structural Semantics
Links are now semantic instances of the indicated associations.
Links are now semantic instances of the indicated associations.
Structural semantics specify how a structural model constrains allowable instance models.
Structural semantics specify how a structural model constrains allowable instance models.
Page 61 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Associations: Behavioral Semantics
• Creating an Order (revised)
• Canceling an Order (revised)
Behavioral semantics specify how a behavioral model changes the state of instances over time.Behavioral semantics specify how a behavioral model changes the state of instances over time.
Page 62 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Creating an Order (revised)
order = new Order (customer, today)
Before After
@createpublic Order (customer: Customer, in datePlaced: Date) { this.datePlaced = datePlaced; this.totalAmount = new Money(dollars=>0, cents=>0);
Customer_Order->add(customer, this);}
@createpublic Order (customer: Customer, in datePlaced: Date) { this.datePlaced = datePlaced; this.totalAmount = new Money(dollars=>0, cents=>0);
Customer_Order->add(customer, this);}
A single action creates a bidirectional link.A single action creates a bidirectional link.
Page 63 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Canceling an Order (revised)
Before Afterorder.cancel()
@destroypublic cancel() { }@destroypublic cancel() { }
A destructor operation is invoked, after which the order object is destroyed.
A destructor operation is invoked, after which the order object is destroyed.
Links in which the destroyed object participates are now also automatically destroyed.
Links in which the destroyed object participates are now also automatically destroyed.
Page 64 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
iv. Asynchronous Communication
• Signals and Receptions
• Classifier Behaviors
• Asynchronous Behavior
Page 65 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Signals and ReceptionsA signal is a classifier whose instances may be communicated asynchronously.
A signal is a classifier whose instances may be communicated asynchronously.
A reception is a declaration of the ability to receive a signal.
A reception is a declaration of the ability to receive a signal.
A signal may have attributes that represent transmittable data.
A signal may have attributes that represent transmittable data.
More than one class can receive the same signal.
More than one class can receive the same signal.
Page 66 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Classifier Behaviors
An active class is one that has a classifier behavior. Only active class may receive signals.
An active class is one that has a classifier behavior. Only active class may receive signals.
A classifier behavior is an autonomous behavior started when an active class is instantiated.
A classifier behavior is an autonomous behavior started when an active class is instantiated.
Page 67 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Asynchronous Behavior
accept (submission: SubmitCharge);card = submission.card;
do { new CreditCardCharge(card, this);
accept (response: ChargeApproved) { declined = false; this.customer.ChargeApproved(response.charge);
} or accept (response: ChargeDeclined) { declined = true; this.customer.ChargeDeclined(response.charge);
}
while (declined);
accept (submission: SubmitCharge);card = submission.card;
do { new CreditCardCharge(card, this);
accept (response: ChargeApproved) { declined = false; this.customer.ChargeApproved(response.charge);
} or accept (response: ChargeDeclined) { declined = true; this.customer.ChargeDeclined(response.charge);
}
while (declined);
The order object accepts a signal to submit a charge.
The order object accepts a signal to submit a charge.
The order object creates a new credit card charge object, which begins its asynchronous behavior.
The order object creates a new credit card charge object, which begins its asynchronous behavior.
Note: UML semantics require a separate action to start the behavior of a new object. However, Alf notation for creating an active class maps to both create and start object behavior actions.
Note: UML semantics require a separate action to start the behavior of a new object. However, Alf notation for creating an active class maps to both create and start object behavior actions.
The order object accepts a signal from the charge object that the charge is approved.
The order object accepts a signal from the charge object that the charge is approved.The order object
sends a signal to the customer that the charge is approved.
The order object sends a signal to the customer that the charge is approved.
Page 68 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
C. Foundational Model Library
• Primitive Behaviors– Boolean Functions– Integer Functions– Unlimited Natural Functions– String Functions
• Basic Input and Output– Common Classes– Channels– Reading and Writing Lines– Hello World
Page 69 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Boolean Functions
Function Signature Description
Or(x: Boolean, y: Boolean): Boolean True if either x or y is true.Post: if x then result = true else result = y endif
Xor(x: Boolean, y: Boolean): Boolean True if either x or y is true, but not both.Post: result = (x Or y) And Not(x And y)
And(x: Boolean, y: Boolean):Boolean True if both x and y are true.Post: if x then result = y else result = true endif
Not(x: Boolean): Boolean True is x is false.Post: if x then result = false else result = true endif
Implies(x: Boolean, y: Boolean): Boolean True if x is false, or if x is true and y is true.Post: result = Not(x) Or (x And y)
ToString(x: Boolean): String Converts x to a String value.Post: if x then result = “true” else result = “false” endif
ToBoolean(x: String): Boolean[0..1] Converts x to a Boolean value.Pre: (lower(x) = “true”) or (lower(x) = “false”)Post: if lower(x) = “true” then result = true else result = false endifNote: The notation “lower(x)” above is not intended to be an invocation of a Foundation Model Library primitive behavior but, rather, is intended to denote that value of the string x with any uppercase letters converted to the corresponding lowercase letters.
Page 70 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Integer FunctionsFunction Signature Description
Neg(x: Integer): Integer The negative value of x.
+(x: Integer, y: Integer): Integer The value of the addition of x and y.
-(x: Integer, y: Integer): Integer The value of the subtraction of x and y.Post: result + y = x
*(x:Integer, y:Integer): Integer The value of the multiplication of x and y.Post:if y < 0 then result =Neg (x * Neg(y))else if y = 0 then result = 0 else result = (x * (y-1)) + xendif endif
Abs(x: Integer): Integer The absolute value of x.Post: if x < 0 then result = Neg(x) else result = x endif
Div(x: Integer, y: Integer): Integer[0..1] The number of times that y fits completely within x.Pre: y<>0Post: if (x * y) >= 0 then((result * y) <= x) And ((result+1) * y) >x)else((Neg(result) * y) <= Neg(x)) And ((Neg(result)+1) * y) > Neg(x))endif
Mod(x: Integer, y: Integer): Integer The result is x modulo y.Post: result = x – (x Div y) * y
Max(x: Integer, y: Integer): Integer The maximum of x and y.Post: if x >= y then result = x else result = y endif
Min(x: Integer, y: Integer): Integer The minimum of x and y.Post: if x <= y then result = x else result = y endif
<(x: Integer, y: Integer): Boolean True if x is less than y.
>(x: Integer, y: Integer): Boolean True if x is greater than y.Post: result = Not(x <= y)
<=(Integer, Integer): Boolean True if x is less than or equal to y.Post: result = (x = y) Or (x < y)
>=(Integer, Integer): Boolean True if x is greater than or equal to y.Post: result = (x = y) Or (x > y)
ToString(x: Integer): String Converts x to a String value.Post: ToInteger(result) = x
ToUnlimitedNatural(x: Integer): UnlimitedNatural[0..1] Converts x to an UnlimitedNatural value.Pre: x >= 0Post: ToInteger(result) = x
ToInteger(x: String): Integer[0..1] Converts x to an Integer value.Pre: x has the form of a legal integer value
Page 71 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Unlimited Natural Functions
Function Signature Description
Max(x: UnlimitedNatural, y: UnlimitedNatural): UnlimitedNatural The maximum of x and y.Post: if x >= y then result = x else result = y endif
Min(x: UnlimitedNatural, y: UnlimitedNatural): UnlimitedNatural The minimum of x and y.Post: if x <= y then result = x else result = y endif
<(x: UnlimitedNatural, y: UnlimitedNatural): Boolean True if x is less than y. Every value other than “unbounded” is less than “unbounded”.
>(x: UnlimitedNatural, y: UnlimitedNatural): Boolean True if x is greater than y.Post: result = Not(x <= y)
<=(UnlimitedNatural, UnlimitedNatural): Boolean True if x is less than or equal to y.Post: result = (x = y) Or (x < y)
>=(UnlimitedNatural, UnlimitedNatural): Boolean True if x is greater than or equal to y.Post: result = (x = y) Or (x > y)
ToString(x: UnlimitedNatural): String Converts x to a String value. The value “unbounded” is represented by the string “*”.Post: ToUnlimitedNatural(result) = x
ToInteger(x: UnlimitedNatural): Integer[0..1] Converts x to an Integer value.Pre: x <> unbounded
ToUnlimitedNatural(x: String): Integer[0..1] Converts x to an Integer value.Pre: (x has the form of a legal integer value) Or (x = “*”)Post:if x = “*” then result = unboundedelse result = ToUnlimitedNatural(ToInteger(x))
Page 72 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
String Functions
Function Signature Description
Concat(x: String, y: String):String The concatenation of x and y.Post:(Size(result) = Size(x) + Size(y)) And(Substring(result, 1, Size(x)) = x) And(Substring(result, Size(x)+1, Size(result)) = y)
Size(x: String):Integer The number of characters in x.
Substring(x: String, lower: Integer, upper: Integer): String[0..1] The substring of x starting at character number lower, up to and including character number upper. Character numbers run from 1 to Size(x).Pre:(1 <= lower) And (lower <= upper) And (upper <= Size(x))
Page 73 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Common Classes
Page 74 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Channels
Page 75 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Reading and Writing Lines
activity ReadLine (out errorStatus: Status[0..1]): String {
return StandardIntputChannel.allInstances().readLine(status);
}
activity ReadLine (out errorStatus: Status[0..1]): String {
return StandardIntputChannel.allInstances().readLine(status);
}
activity WriteLine (in value: String, out errorStatus: Status[0..1]) {
StandardOutputChannel.allInstances().writeLine(result, status);
}
activity WriteLine (in value: String, out errorStatus: Status[0..1]) {
StandardOutputChannel.allInstances().writeLine(result, status);
}
Page 76 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Hello World
activity Hello() {
WriteLine("Hello World!"); }
activity Hello() {
WriteLine("Hello World!"); }
Page 77 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
III. Extensions for Executable SysML
Objectives
• To understand how the semantic specification for a UML profile like SysML can be built on the fUML foundation.
• To provide an overview of some of the key semantic extensions required for SysML.
Page 78 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
III. Extensions for Executable SysML
Topics
A. SysML Semantics
B. Streaming
C. Timing
D. Blocks and Partitions
E. State Machines
Page 79 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Composite Structure
Semantics
Composite Structure
Semantics
Complete Activity Model
Semantics
Complete Activity Model
Semantics
A. SysML Semantics
State Machine Semantics
State Machine Semantics
Non-Executable
Model Semantics
Non-Executable
Model Semantics
Interaction Model
Semantics
Interaction Model
Semantics
Foundational SemanticsFoundational Semantics
SysML is base on a UML subset that does not include all of fUML (e.g., it does not include structured activity
nodes), but it includes capabilities not in fUML (e.g., composite structure,
state machines, streaming, etc.)
SysML is base on a UML subset that does not include all of fUML (e.g., it does not include structured activity
nodes), but it includes capabilities not in fUML (e.g., composite structure,
state machines, streaming, etc.)
UML for SysML
SysML also defines a profile consisting of stereotypes used to tag
standard UML model elements to give them SysML-specialized meaning.
SysML also defines a profile consisting of stereotypes used to tag
standard UML model elements to give them SysML-specialized meaning.
Page 80 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Building on the Foundation
There are two main ways to build on the semantic foundation provided by fUML.
1. Translate higher-level constructs into equivalent fUML models.• For example, provide semantics for state machines by specifying
how to translate them into equivalent fUML activity models.
2. Extend the fUML execution model to directly specify the semantics of higher-level constructs.• For example, extend the fUML specification for behavior invocation
and action pins to define the semantics of stream parameters.
Page 81 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
B. Streaming
• Streaming is a property of a behavior parameter that gives the behavior access to from its invoker while the behavior is executing.
• Multiple values may arrive through a streaming input parameter during a single behavior execution, not just at the beginning.
• Multiple values may be posted to a streaming output parameter during a single behavior execution, not just at the end.
Page 82 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Based on the sample activity in Annex B.4.8.1 of the SysML Specification
…[execute] Activity ProvidePower completed.[addToken] node = out drivePower[addToken] node = out drivePower[addToken] node = out transModeCmd[receiveOffer] node = drivePower[fire] Output activity parameter node drivePower...[addToken] node = drivePower[removeToken] node = out drivePower[addToken] node = drivePower[removeToken] node = out drivePower[receiveOffer] node = transModeCmd[fire] Output activity parameter node transModeCmd...[addToken] node = transModeCmd[removeToken] node = out transModeCmd
Execution without Streaming[execute] Activity Accelerate...[run] Node MeasureVehicleConditions is enabled.[run] Node PushAccelerator is enabled.[run] Sending offer to node MeasureVehicleConditions.[fire] Action MeasureVehicleConditions...[execute] Activity MeasureVehicleConditions...…[execute] Activity MeasureVehicleConditions completed.[addToken] node = out vehCond[receiveOffer] node = ProvidePower[run] Sending offer to node PushAccelerator.[receiveOffer] node = PushAccelerator[fire] Action PushAccelerator...[addToken] node = out accelPosition[receiveOffer] node = ProvidePower[addToken] node = in accelPosition[removeToken] node = out accelPosition[addToken] node = in vehCond[removeToken] node = out vehCond[fire] Action ProvidePower...[execute] Activity ProvidePower...
output parameter 'drivePower' has 2 value(s)value[0] = Reference to (Object_@130be8c: GasPower throttle = 6)value[1] = Reference to (Object_@12df081: ElecPower current = 3)output parameter 'transModeCmd' has 1 value(s)value[0] = Reference to (Object_@108d3eb: TransmissionModeCommand)
The execution of the activity is shown by its execution trace.The execution of the activity is shown by its execution trace.
Page 83 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Execution with streaming…[fire] Action ProvidePower...[execute] Activity ProvidePower...…[addToken] node = out drivePower[receiveOffer] node = drivePower[fire] Output activity parameter node drivePower...[addToken] node = drivePower[removeToken] node = out drivePoweroutput parameter 'drivePower' posts 1 value(s)value[0] = Reference to
(Object_@130be8c: GasPower throttle = 6)
…[addToken] node = out transModeCmd[receiveOffer] node = transModeCmd[fire] Output activity parameter node transModeCmd...[addToken] node = transModeCmd[removeToken] node = out transModeCmdoutput parameter 'transModeCmd' posts 1 value(s)value[0] = Reference to
(Object_@108d3eb: TransmissionModeCommand)…
…[addToken] node = out drivePower[receiveOffer] node = drivePower[fire] Output activity parameter node drivePower...[addToken] node = drivePower[removeToken] node = out drivePoweroutput parameter 'drivePower' posts 1 value(s)value[0] = Reference to (Object_@12df081: ElecPower current = 3)
ProvidePower is still executing at this point…
Page 84 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
C. Timing
• Timing constraints model requirements for how behaviors execute over time.
• A time constraint specifies that some event is required to happen within a given time interval.
• A duration constraint specifies that the duration between two events must be within a given duration interval.
Page 85 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Execution with timing
[fire] Action MeasureVehicleConditions…[execute] Activity MeasureVehicleConditions……[addToken] node = out vehCond…[fire] Action PushAccelerator...[execute] Activity PushAccelerator……[addToken] node = out accelPosition…[fire] Action ProvidePower...[execute] Activity ProvidePower...…[addToken] node = out drivePower
2 sec
1 sec
3 sec
Duration
t = 0
t = 2
t = 2
t = 3
t = 3
t = 6
Sequential execution
t = 0
t = 2
t = 0
t = 1
t = 2
t = 5
Parallel execution A duration constraint on the duration between the firing of the action and the placing of tokens on its output pin.
A duration constraint on the duration between the firing of the action and the placing of tokens on its output pin.
Page 86 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Timing: Additional Considerations
• Duration ranges– The constraint on a duration may be that it is within a certain range,
rather than exactly equal to a given value.
• Probability distribution– A Monte Carlo simulation may select the duration of an action
execution on each run to be a random value within a constrained duration range, with some given probability distribution.
• Local clocks– The flow of time is modeled by adding the concept of a clock to the
fUML execution model.– However, rather than a single global clock, there maybe a number of
local clocks that define time coordinates that flow at different rates than each other.
– The base fUML execution semantics do not depend on any specific model of time.
Page 87 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
D. Blocks and Partitions
• In SysML, a block is a modular unit of system description.– SysML blocks are based on UML classes, possibly with composite
structure.
• An activity partition is a grouping of nodes within an activity that share some common characteristic.
• An activity may be partitioned such that the behavior grouped in each partition is allocated to a specific block, which is responsible for carrying out that behavior.
• The overall behavior modeled by the full activity then emerges from the interaction between the blocks to which various partitions of the activity have been allocated.
Page 88 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Allocating Activities to BlocksFlows across partitions define interfaces between blocks
Page 89 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Allocated Block StructureAn internal block diagram shows the composite structure of the Power Subsystem.
An internal block diagram shows the composite structure of the Power Subsystem. Each component block
has behavior as allocated from the activity model.
Each component block has behavior as allocated from the activity model.
Connections between flow ports on the blocks are derived from flows that cross partition boundaries.
Connections between flow ports on the blocks are derived from flows that cross partition boundaries.
Page 90 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Blocks and Partitions: Considerations
• Execution semantics for SysML blocks and ports
• The semantics of allocation of activities to blocks
• The formal relationship between activity object flows and block flow ports
Page 91 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
E. State Machines
• A state machine abstracts system behavior into a finite number of states and the transitions between those states.
• Behavior may be triggered as the effect of a transition or on entry, on exit or while in a state.
• State machine semantics are described in detail in the UML Superstructure specification, but this is not formalized.
• State machine semantics have been formalized by Harel and others, but these formalizations have not been standardized for UML.
• Basic state machine semantics can also be understood by translations to equivalent activity semantics, as formalized in fUML.
Page 92 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Appendix: Specifying Execution Semantics
• Background– Models– Metamodeling and Semantics– Denotational Mapping
• Approach– Semantics of Values– Semantics of Behavior– Execution Semantics and Base Semantics– Execution Environment
Page 93 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
ModelsA model makes statements in a modeling language about a system under study.A model makes statements in a modeling language about a system under study.
UML Instance Model
UML Class Model
First order statements
“There is a person whose name is Jack.”
”There is a house. The person is the owner of the house.”
First order statements
“There is a person whose name is Jack.”
”There is a house. The person is the owner of the house.”
Second order statements
“Every person has a name.”
“Some people own houses.”
Second order statements
“Every person has a name.”
“Some people own houses.”
Page 94 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Metamodeling and Semantics
X : anX
anXX.java "instance of"
Class InstanceSpecification
: InstanceSpecification: Class +classifier
"interpretation"
<<instanceOf>>
"interpretation"
<<instanceOf>>
"interpretation"
"interpretation"
<<instanceOf>>
+classifier
M1
M0
M2
M1
“interpretation”
“interpretation”"interpretation"
“representation"
References• Ed Seidewitz, “What Models Mean,” IEEE Software, September/October 2003.• Ed Seidewitz, “What do models mean?”, OMG document ad/03-03-31.
Page 95 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Denotational Mapping
evaluate(specification: ValueSpecification): Value
Abstract Syntax Element
(Representation)
Abstract Syntax Element
(Representation)
Semantic Model Element
(Interpretation)
Semantic Model Element
(Interpretation)
Page 96 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Abstract Syntax: Value Specifications
Page 97 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Semantics: Values
Page 98 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Representation: Instance Model
Page 99 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Interpretation: Instance Model
j = evaluate(v)j = evaluate(v)
Page 100 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Semantics: Extensional Values
There are concepts in the semantic model that
have no explicit representation in the
abstract syntax.
There are concepts in the semantic model that
have no explicit representation in the
abstract syntax.
Page 101 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Abstract Syntax/Semantics: Behavior
Page 102 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Abstract Syntax: Activities
Page 103 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Semantics: Activities
Additional semantic concepts have
specifically to do with dynamic behavior.
Additional semantic concepts have
specifically to do with dynamic behavior.
Page 104 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Model: Simple Activity
Page 105 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Representation: Simple Activity
Page 106 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Interpretation: Simple Activity Execution (1)
Page 107 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Interpretation: Simple Activity Execution (2)
Page 108 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Execution Semantics and Base Semantics
(forall (n a xa f xn)(if (and (ExecutableNode n)
(buml:activity n a)(classifies a xa f)(property-value xa n xn f)
(ipc:subactivity_occurrence-neq xn xa))
(forall (n a xal xa2 xn)(if (and (ExecutableNode n)
(buml:activity n a)(classifies a xa1 f)(classified a xa2 f)(property-value xa1 n xn f)(property-value xa2 n xn f)
(= (psl:root occ xa1) (psl:root occ xa2))))
(forall (n a xa f xn)(if (and (ExecutableNode n)
(buml:activity n a)(classifies a xa f)(property-value xa n xn f)
(ipc:subactivity_occurrence-neq xn xa))
(forall (n a xal xa2 xn)(if (and (ExecutableNode n)
(buml:activity n a)(classifies a xa1 f)(classified a xa2 f)(property-value xa1 n xn f)(property-value xa2 n xn f)
(= (psl:root occ xa1) (psl:root occ xa2))))
Execution Semantics(Operational Specification)
Base Semantics(Axiomatic Specification)
• Foundational UML (fUML) semantics are specified operationally as a UML Model written in Base UML (bUML).
• Base UML semantics are (to be) specified axiomatically using PSL.
• Foundational UML (fUML) semantics are specified operationally as a UML Model written in Base UML (bUML).
• Base UML semantics are (to be) specified axiomatically using PSL.
Page 109 Copyright © 2010 Data Access Technologies, Inc.
(Model Driven Solutions)February 2010
Execution Environment
• Locus– Manages extents– Provides pre-instantiated discoverable services
• Executor– Evaluates value specifications– Executes behaviors (synchronously)– Starts behaviors or active objects (asynchronously)
• Execution Factory– Creates visitor objects– Registers strategies– Registers primitive types and primitive behaviors