Reusable specification templates for defining dynamic semantics - - PowerPoint PPT Presentation

Reusable specification templates

Ulyana Tikhonova

ulyana.tikhonova@constelle.net

for defining dynamic semantics

• f DSLs

Mark van den Brand Tim Willemse

Domain Specific Languages (DSLs)

2

3

Why to define dynamic semantics?

4

• To understand the dynamic semantics
• To analyze the dynamic semantics
• To validate the dynamic semantics
• To understand and debug

DSL programs

Precise and executable definition

DSL developer DSL user

What is dynamic semantics?

6

What is dynamic semantics?

7

1 2 3 4 5

What is dynamic semantics?

8

1 2 3 4 5

• To understand the dynamic semantics
• To analyze the dynamic semantics
• To validate the dynamic semantics
• To understand and debug DSL programs

Dynamic semantics as requirements vs. dynamic semantics as an actual implementation

9

• To understand the dynamic semantics
• To analyze the dynamic semantics
• To validate the dynamic semantics
• To understand and debug DSL programs

10

Specification templates

Defining dynamic semantics of programming languages

11

Semantic mapping Semantic domain

• Variables/memory
• Control flow
• Branching

Language constructs (statements)

Defining dynamic semantics of domain specific languages

12

Semantic mapping Semantic domain

• Variables/memory
• Control flow
• Branching

DSL constructs (statements)

Defining dynamic semantics of domain specific languages

13

Semantic mapping Semantic domain

• Variables/memory
• Control flow
• Branching

DSL constructs (statements) Semantic domain Semantic mapping

• Architecture layers
• Design patterns
• Synchronization protocols

Specification templates

Defining dynamic semantics of domain specific languages

14

Semantic mapping Semantic domain

• Variables/memory
• Control flow
• Branching

DSL constructs (statements) Semantic domain Semantic mapping

• Architecture layers
• Design patterns
• Synchronization protocols

Specification templates

16

MACHINE queue_machine SEES queue_context VARIABLES queue INVARIANTS inv1: queue ∈ ℕ ⇸ ElementType EVENTS INITIALISATION ≙ act1: queue ≔ ∅ END enqueue≙ ANY element, index WHERE grd1: element ∈ ElementType grd2: index ↦ element ∈ queue grd3: ∀i·i ∈ dom(queue) ⇒ index ≤ i THEN act1: queue ≔ queue ∖ {index ↦ element} END dequeue≙ ANY element, index WHERE grd1: element ∈ ElementType grd2: index ∈ ℕ grd3: queue ≠ ∅ ⇒ (∀i·i ∈ dom(queue) ⇒ index > i) grd4: {index ↦ element} ∈ ℕ ⇸ ElementType grd5: index ∉ dom(queue) THEN act2: queue ≔ queue ∪ {index ↦ element} END END

MyType MyType MyType MyType

17

MACHINE queue_machine SEES queue_context VARIABLES queue INVARIANTS inv1: queue ∈ ℕ ⇸ ElementType EVENTS INITIALISATION ≙ act1: queue ≔ ∅ END dequeue≙ ANY element, index WHERE grd1: element ∈ ElementType grd2: index ∈ ℕ grd3: queue ≠ ∅ ⇒ (∀i·i ∈ dom(queue) ⇒ index > i) grd4: {index ↦ element} ∈ ℕ ⇸ ElementType grd5: index ∉ dom(queue) THEN act2: queue ≔ queue ∪ {index ↦ element} enqueue≙ ANY element, index WHERE grd1: element ∈ ElementType grd2: index ↦ element ∈ queue grd3: ∀i·i ∈ dom(queue) ⇒ index ≤ i THEN act1: queue ≔ queue ∖ {index ↦ element} END

Logical action Subsystem action

• ccurrence

Subsystem Subsystem action *

bo dy

1

typ e

1

typ e

* Order Scan Data Node Data type Dataflow Data Transformation

1

typ e

*

1. .*

* *

0. .1 1 1

t

• fro

m

1 1

t

• fro

m

* * * * *

inp ut

• utput

inp ut

• utput

DSL metamodel Event-B specification

Structural interface

Substitution

Structural interface

Constelle

MyType MyType MyType MyType

Defining dynamic semantics of domain specific languages

18

Semantic mapping Semantic domain

• Variables/memory
• Control flow
• Branching

DSL constructs (statements) Semantic domain Semantic mapping

• Architecture layers
• Design patterns
• Synchronization protocols

Specification templates

Mapping DSL constructs to specification templates

19

1 2 3 4 5

20

VARIABLES curr_job, curr_la, la_input, ssa_output INVARIANTS la_input ∈ ℕ ⇸ LogicalActions ssa_output ∈ ℕ ⇸ SSActions curr_job ∈ ℙ(SSAOccurences) curr_la ∈ LogicalActions EVENTS Initialisation curr_la :∈ LogicalActions curr_job ≔  la_input ≔  ssa_output ≔  request_la (la, n) where la ∈ LogicalActions curr_job =  n ∈ ℕ la_input ≠  ⇒ ∀ i · i ∈ dom(la_input) ⇒ n > i then curr_job ≔ dom(LALabelDef(la)) curr_la ≔ la la_input ≔ la_input ∪ { n ↦ la } request_ssa (ssaction, occurence) where

• ccurence ∈ curr_job
• ccurence ↦ ssaction ∈ LALabelDef (curr_la)

then curr_job ≔ curr_job \ {occurence} execute_ssa (ssaction, n) where ssaction ∈ SSActions n ∈ ℕ ssa_output ≠  ⇒ ∀ i · i ∈ dom(ssa_output) ⇒ n > i then ssa_output ≔ ssa_output ∪ { n ↦ ssaction } END

Aspect Oriented Programming: cross cutting concerns

21

Queue enqueue dequeue Listener subscribe notify Partial Order init_partial_order is_max_element remove_element

• peration1

subscribe init_partial_order

• peration2

notify enqueue is_max_element

• peration3

dequeue remove_element

Aspect Oriented Programming: composing cross cutting concerns

22

Listener Queue Partial Order

• peration1

subscribe init_partial_order

• peration2

notify enqueue is_max_element

• peration3

dequeue remove_element Specializations of specification templates from the library

Constelle

23

Substitution of template (static) parameters with DSL types Operations

• f the DSL

dynamic semantics Aspects = specification templates

Constelle

24

Meta-model DSL model Event-B specification templates Event-B specification

DSL/Ecore Event-B/Rodin

Specification templates Constelle definition Constelle-to-Event-B

Future work

25

• Scalability and applicability of the proposed method

– Validated on three different DSLs

• A library of specification templates

– State machine, communication channels, data typing, design patterns (Listener), data structures (Queue)

• Specification templates for bridging

technological diversity

– Source code, visualization templates, formal notations

More details: SoSyM paper & PhD thesis

26