The e Hi High-Level el Variability Language: e: an ontological - - PowerPoint PPT Presentation

The e Hi High-Level el Variability Language: e: an ontological approach

Angela Villota, Raúl Mazo and Camille Salinesi

Context

Differences

• Style: orthogonal
• Targets
• Structure
• Way to be used

Similarities

• Syntax
• Semantics

Context

Translational Semantics

Map models in propositional formulas or constraint satisfaction problems to perform reasoning tasks.

Constraints

Why don’t we use a constraint-based language to model variability?

High-Level Constraint Language

• Djebbi et al. (2009)
• Mazo et al. (2011)
• Dumitrescu et al. (2014)
• Muñoz et al. (2015)
• Villota et al. (2018)

Context

REFAS: variability, goals, soft-goals, claims Constraint graphs Feature models

The Ontological Path

• Comparison with other PLMs languages.
• User experience.
• A more formal approach?

Ontological approach

• To evaluate the expressiveness of the intermediate language
• To conceptualize and structure knowledge about variability modeling
• To define variability constructs and language characteristics

Legend

Asadi's et al. ontology Knowledge sources Completeness and clarity criteria Theory of ontological expressiveness Variability Modeling glossary HLVL Ontological analysis

1

Conceptualization

2

Refinement and synthesis

3

Process input/

• utput

Can we model variability comprehensively using HLCL?

V

V1: Medium- class car

VP Mandatory Optional V Variant Mandatory Optional [min, max]

Alternative

Requires Excludes VP

VP

VP1: Type

• f vehicle

VP

VP2: Activation

VP VP3:

Confort functions

VP

VP4: Other signs

VP

VP5: Prohibition signs

V

V2: Upper- class car

V

V3: Small truck(3, 5t)

V

V4: Big truck(7, 5t) 1, 1

V

V5: Switchable

V

V6: Durable 1, 1

V

V7: No stopping warning

V

V8: Overspeed warning

V

V11: No vehicles

V

V9: Road w/right

• f way start

V

V12: No cars

V

V10: City limit

What to buy? (name: scope; expected val 1:1): {"assemble yourself", "complete suite"}) Which tools? (name: tools; expected val 1:3): {"CW", "DK", "PK"}) Include glossary? (name: glossary; expected val: bool) Default resolution? (name: resolution; expected val 1:1): {"800x600", }) Width? (name: width; expected val: number)

• Decision effect

Validity Cond.

• Visibility Cond.
• Decision

Parking Assistant System Processor Memory

Name: size Domain: integers Values: [2, 4, 8, 16, 32] Name: cores Domain: integers Value: 1..7

Excludes Requires Mandatory Optional OR

Alternative

feature Attribute

Speed sensor Sensors Position sensor Feedback Visual Audio

[1, 2]

Vibration

Variability unit Variability relation Variant

Feature-oriented Decision-oriented VP-oriented Constraint-oriented

Variability modeling languages

not completely different, not completely the same

Va Variability co concepts

The High-Level Variability Language

• Rich textual language to model variability.
• Syntax that resembles programming languages, to make it human

readable

• HLVL is a language capable of supporting concepts of many variability
• languages. Why just features?

Would you use it?

• Would you leave your modeling tools for an HLVL-tool?

The High-Level Variability Language

Payment E-shop Customer type Connection Insecure Secure PayPal Card Gift Credit Regular Sporadic Constraints Regular => Customer profile ~(Insecure /\ Credit)

Name: confidentiality Domain: integers Value: 5 Name: confidentiality Domain: integers Value: 3..5

Excludes Requires Mandatory Optional OR

Alternative

feature Attribute

Debit AppServer Implementation Machines

[1, 5] [2, 10] [0, 1]

The High-Level Variability Language

V

V1: Medium- class car

VP Mandatory Optional V Variant Mandatory Optional [min, max]

Alternative

Requires Excludes VP

VP

VP1: Type

• f vehicle

VP

VP2: Activation

VP VP3:

Confort functions

VP

VP4: Other signs

VP

VP5: Prohibition signs

V

V2: Upper- class car

V

V3: Small truck(3, 5t)

V

V4: Big truck(7, 5t) 1, 1

V

V5: Switchable

V

V6: Durable 1, 1

V

V7: No stopping warning

V

V8: Overspeed warning

V

V11: No vehicles

V

V9: Road w/right

• f way start

V

V12: No cars

V

V10: City limit

The High-Level Variability Language

What to buy? (name: scope; expected val 1:1): {"assemble yourself", "complete suite"}) Which tools? (name: tools; expected val 1:3): {"CW", "DK", "PK"}) Include glossary? (name: glossary; expected val: bool) Default resolution? (name: resolution; expected val 1:1): {"800x600", }) Width? (name: width; expected val: number)

• Decision effect

Validity Cond.

• Visibility Cond.
• Decision

What’s next?

Domain engineers model variability using their preferred language/tool 2 Then, models are transformed to produce their HLVL representation

model mobilePhone elements: boolean calls boolean GPS boolean screen boolean basicScreen boolean hrScreen boolean camera relations: R0: coreElements(calls, screen) R1: mutex(GPS, basicScreen) R2: implies(camera, hrScreen)

HLVL

The transformation process is shorter using our HLVL variability modelling editor. 3 Variability models represented in HLVL are ready for the reasoning process H L H L V L V L 1 1 2 3 4 HLVL models are transformed into constraint programs in HLCL 5 Models and reasoning questions are processed to choose the best solving tool and algorithm for answering each question HL HLCL H L H L V L V L 4 5 6 6 The solving tool is executed, the output is processed, and the questions about the variability model are answered.

HLCL

% Autogenerated code from % the Coffee framework % Variables from elements defs. var 0..1: calls; var 0..1: GPS; var 0..1: screen; var 0..1: basicScreen; var 0..1: hrScreen; var 0..1: camera; % Variables and constraints constraint calls = 1; constraint screen = 1; constraint GPS + basicScreen <1; constraint hrScreen < camera; % Solving parameters solve satisfy;

Coffee

Angela Villota

Contact: angievig@gmail.com

Thank you!