Software Variability Management

Software variability management
Xavier Devroey <x.d.m.devroey@tudelft.nl>
Office 4.W.740 (4th floor, West side, VMB building 28)

Variability in platforms
Operating System:
Database Management System:

Variability in software
- Bundles
https://www.eclipse.org

- Bundles
- Command line parameters
OVERVIEW: clang LLVM compiler
USAGE: clang [options] <inputs>
OPTIONS:
-### Print (but do not run) the commands to run for this compilation
--analyzer-output <value>
Static analyzer report output format (html|plist|plist-multi-file|plist-html|text).
--analyze Run the static analyzer
-arcmt-migrate-emit-errors
Emit ARC errors even if the migrator can fix them
-arcmt-migrate-report-output <value>
Output path for the plist report
-B <dir> Add <dir> to search path for binaries and object files used implicitly
-CC Include comments from within macros in preprocessed output
-cfguard Emit tables required for Windows Control Flow Guard.
-cl-denorms-are-zero OpenCL only. Allow denormals to be flushed to zero.
-cl-fast-relaxed-math OpenCL only. Sets -cl-finite-math-only and -cl-unsafe-math-optimizations, and defines __FAST_RELAXED_MATH__.
-cl-finite-math-only OpenCL only. Allow floating-point optimizations that assume arguments and results are not NaNs or +-Inf.
-cl-fp32-correctly-rounded-divide-sqrt
OpenCL only. Specify that single precision floating-point divide and sqrt used in the program source are correctly rounded.
-cl-kernel-arg-info OpenCL only. Generate kernel argument metadata.
-cl-mad-enable OpenCL only. Allow use of less precise MAD computations in the generated binary.
-cl-no-signed-zeros OpenCL only. Allow use of less precise no signed zeros computations in the generated binary.
-cl-opt-disable OpenCL only. This option disables all optimizations. By default optimizations are enabled.
-cl-single-precision-constant
OpenCL only. Treat double precision floating-point constant as single precision constant.
-cl-std=<value> OpenCL language standard to compile for.
-cl-strict-aliasing OpenCL only. This option is added for compatibility with OpenCL 1.0.
-cl-uniform-work-group-size
OpenCL only. Defines that the global work-size be a multiple of the work-group size specified to clEnqueueNDRangeKernel
-cl-unsafe-math-optimizations
OpenCL only. Allow unsafe floating-point optimizations. Also implies -cl-no-signed-zeros and -cl-mad-enable.
--config <value> Specifies configuration file
--cuda-compile-host-device
Compile CUDA code for both host and device (default). Has no effect on non-CUDA compilations.
--cuda-device-only Compile CUDA code for device only
--cuda-gpu-arch=<value> CUDA GPU architecture (e.g. sm_35). May be specified more than once.
--cuda-host-only Compile CUDA code for host only. Has no effect on non-CUDA compilations.
--cuda-include-ptx=<value>
Include PTX for the follwing GPU architecture (e.g. sm_35) or 'all'. May be specified more than once.
--cuda-noopt-device-debug
Enable device-side debug info generation. Disables ptxas optimizations.
--cuda-path-ignore-env Ignore environment variables to detect CUDA installation
--cuda-path=<value> CUDA installation path
-cxx-isystem <directory>
Add directory to the C++ SYSTEM include search path
-C Include comments in preprocessed output
-c Only run preprocess, compile, and assemble steps
-dD Print macro definitions in -E mode in addition to normal output
-dependency-dot <value> Filename to write DOT-formatted header dependencies to
-dependency-file <value>
Filename (or -) to write dependency output to
-dI Print include directives in -E mode in addition to normal output
-dM Print macro definitions in -E mode instead of normal output
-D <macro>=<value> Define <macro> to <value> (or 1 if <value> omitted)
-emit-ast Emit Clang AST files for source inputs
-emit-llvm Use the LLVM representation for assembler and object files
-enable-trivial-auto-var-init-zero-knowing-it-will-be-removed-from-clang<value>
Trivial automatic variable initialization to zero is only here for benchmarks, it'll eventually be removed, and I'm OK with that because I'm only using it to
benchmark
-E Only run the preprocessor
-faddrsig Emit an address-significance table
-faligned-allocation Enable C++17 aligned allocation functions
-fallow-editor-placeholders
Treat editor placeholders as valid source code
-fansi-escape-codes Use ANSI escape codes for diagnostics
-fapinotes-cache-path=<directory>
Does nothing; API notes are no longer cached separately from modules
-fapinotes-modules Enable module-based external API notes support
gcc -help

- Bundles
- Plugins

- Bundles
- Plugins
- Conﬁguration ﬁles
<profiles>
<profile>



<id>release</id>
<build>
<plugins>
[…]
</plugins>
</build>
</profile>
</profiles>
https://github.com/STAMP-project/botsing/blob/master/pom.xml

- Bundles
- Plugins
- Conﬁguration ﬁles
- Microservices
☛

software variability is the ability of
a software system or artefact to be efficiently
extended, changed, customised or configured
for use in a particular context.
Svahnberg, M., et al. (2005) ‘A taxonomy of variability realization techniques’, Software - Practice and
Experience, 35(8), pp. 705–754.

Does your application have variability?
Yes, most probably
• What kind of variability?
• Hardware
• Platform
• Software
• Bundels
• Plugins
• Command line options
• Configuration files
• Microservices
• …

33
Options
0
A unique product for every person on this planet
(8,589,934,592 products)

320
More products than the
estimated atoms in the universe
(2.135987e+96 products)
33
Options
0

How to…
…develop?
…test?
…maintain?
…define requirements?

Clone and Own
(Variability 101)

Problem: how to manage the evolution of a large number of copies?

Software Variability Management

•Fast
•Cheap (at first)
•Limited
•Manage (large number of) copies
•Maintenance cost at each evolution
•Inconsistent evolution of copies

Software Product Line
(Variability level over 9000!)

A product is built by systematically combining 
commonalities, common to all products, and  
variabilities, specific to some products.
Features
A feature is a characteristic
or end-user-visible behaviour
of a software system.
Apel, S. et al. (2013) Feature-Oriented Software Product Lines. Springer.

The software product line engineering framework
Commonalities and
variabilities 
definition and realisation
Product derivation
Pohl, K., et al. (2005) Software product line engineering: foundations, principles, and techniques. Springer.

Example: SAP, an Enterprise Resource Planning system (ERP)
https://retool.com/blog/erp-for-engineers/

Integrated management of business processes and operational data
Database
Core
Customer-
relationship
management
(CRM)
API
Sales Accounting HR …

ERPs in numbers
- SAP was founded in 1972
- Designed to be extensible from the beginning
- Every year (worldwide)…
- companies spend $41B on ERPs
- SAP alone sells for $25B
- SAP systems manage 77% of the world’s transaction revenue, and
78% of the world’s food

6,88832033
Options
0 >8,000
A basic SAP installation has 20,000 database tables,
3,000 are configuration tables
SAP requires ~8,000 configuration decisions at setup
☛ SAP Configuration Specialist is an actual job title!

ERPs in numbers
- Switching to an ERP
- Costs $100 million to $500 million
- $30 million in software license fees,
- $200 million in consulting fees,
- millions for hardware,
- and millions to train managers and employees.
- Takes from four to six years
- Examples
- Cisco’s ERP implementation took 9 months and $15 million (best case)
- Dow Chemical’s implementation took 8 years and $1 billion
- the U.S. navy spent $1 billion on four different ERP projects that all
failed

Pohl, K., et al. (2005) Software product line engineering: foundations, principles, and techniques. Springer.
- Building the ERP platform
- Core
- API
- Modules
- Configuration workflow
- Configuration
- Deployment
- Training

Example: Enterprise resource planning systems (ERPs)
Identification of
business processes
Configuration
Customisation and
extension
Data migration
Redefinition of business
processes

Hall of Fame Software Product Line Conference (SPLC)
http://splc.net/hall-of-fame/
- Boeing
- Bosch Group
- CelsiusTech Systems AB
- Cummins, Inc.
- Danfoss
- Ericsson AXE
- FISCAN
- General Motors Powertrain
(GMPT)
- Hewlett Packard
- HomeAway
- Lockheed Martin
- LSI Logic
- Lucent
- Market Maker
- Nokia
- Philips (Medical Systems)
- Philips (Software for Television
Sets)
- Philips (Telecommunication
Switching System)
- Salion, Inc.
- Siemens (syngo.via)
- Toshiba
- U.S. Army Live Traning
Transformation
- U.S. Naval Research
Laboratory

•Tailor-made
• For individual customers
•Reduced costs
• Reusable parts can be combined in different ways
•Improved quality
• Parts can be standardized and checked in isolation
and reused and tested in multiple products.
•Time to market
• building on top of existing parts is much faster than
developing it entirely from scratch

4833 320 6,888
Options
0
Halin, A., et al. 2018. Test them all, is it worth it? Assessing configuration sampling on the JHipster Web development stack. EMSE.

A feature model documents
the features of a product line
and their relationships.

Vocabulary
Feature = Option
Product = Configuration = Variant

• Tree-like structure (directed acyclic graph).
• A feature is decomposed in sub-features
Feature diagram
• And decomposition: if the parent is selected then all sub-features
must be selected too
• Or decomposition: if the parent is selected then at least one sub-
features must be selected too
P
C1 C2 C3
Valid product:
{P, C1, C2, C3}
P
C1 C2 C3
Valid products:
{P, C1} {P, C2} {P, C3}
{P, C1, C2} {P, C2, C3} {P, C1, C3}
{P, C1, C2, C3}

• Tree-like structure (directed acyclic graph).
• A feature is decomposed in sub- features
• Xor decomposition: if the parent is selected then exactly one sub-
features must be selected too
P
C1 C2 C3
Valid products:
{P, C1}
{P, C2}
{P, C3}
• Mandatory feature: if the
parent is selected then the
feature must be selected
• Optional feature: if the parent
is selected then the feature
must be selected
P
C
P
C
Valid product:
{P, C}
Valid products:
{P}
{P, C}
Feature diagram

• Cross-tree constraints are expressed as boolean expressions
over the features.
• The semantics of a feature model corresponds to all the valid
products allowed by the feature model.
P
C1 C2 C3
(C1 ∧ C2) C3
Valid products:
{P, C1} {P, C2} {P, C3}
{P, C1, C2} {P, C2, C3} {P, C1, C3}
{P, C1, C2, C3}
{{P, C1} {P, C2} {P, C3}
{P, C1, C2} {P, C2, C3} {P, C1, C3}
{P, C1, C2, C3}}
P
C1 C2 C3
[| |]=
Feature diagram

Constraints and where to find them

Source: Configurator
https://start.jhipster.tech/
OR
decomposition
XOR
decomposition
AND
decomposition
Optional feature

Source: Documentation
Which type of application would you like to
create?
You can either use:
• Monolithic application […]
• Microservice application […]
• Microservice gateway […]
http://www.jhipster.tech/creating-an-app/
Which type of database would you like to use?
You can choose between:
• No database (only available when using a
microservice application)
• An SQL database (H2, MySQL, MariaDB,
PostgreSQL,MySQL, Oracle), which you will access
with Spring Data JPA
• MongoDB
• Cassandra
Social login (Google, Facebook, Twitter)
This option is only available if you selected an SQL
or MongoDB database.
XOR
decomposition
Additional
constraints
SocialLogin ⇒ (SQL ∨ MongDB)
XOR
decomposition
Optional feature
Additional
constraints
¬Database ⇒ MicroserviceApplication
SQL ⇒ SpringDataJPA

Source: Source code
public void invalidateSession(@PathVariable String series) throws UnsupportedEncodingException {
String decodedSeries = URLDecoder.decode(series, "UTF-8");
SecurityUtils.getCurrentUserLogin()
.flatMap(userRepository::findOneByLogin)
.ifPresent(u ->
persistentTokenRepository.findByUser(u).stream()
.filter(persistentToken -> StringUtils.equals(persistentToken.getSeries(),
decodedSeries))
<%_ if (databaseType === 'sql' || databaseType === 'mongodb') { _%>
.findAny().ifPresent(t -> persistentTokenRepository.deleteById(decodedSeries)));
<%_ } else if (databaseType === 'couchbase') { _%>
.findAny().ifPresent(t -> persistentTokenRepository.deleteBySeries(decodedSeries)));
<%_ } else { _%>
.findAny().ifPresent(persistentTokenRepository::delete));
<%_ } _%>
}<% } %>
https://github.com/jhipster/generator-jhipster/blob/master/generators/server/templates/src/main/java/package/web/rest/AccountResource.java.ejs
OR
decomposition
Mandatory
features

http://www.featureide.com

Halin, A., et al. 2018. Test them all, is it worth it? Assessing configuration sampling on the JHipster Web development stack. EMSE.
https://github.com/xdevroey/jhipster-dataset/tree/master/v3.6.1/featuremodel
JHipster v3.6.1

JHipster-App ∧ (Building ∨ ¬Maven) ∧ (Building ∨ ¬Gradle) ∧ (Maven ∨ Gradle
∨ ¬Building) ∧ (¬Maven ∨ ¬Gradle) ∧ (Server ∨ ¬Database) ∧ (Server ∨
¬HibernateCache) ∧ (Server ∨ ¬Building) ∧ (Building ∨ ¬Server) ∧ (JHipster-
App ∨ ¬Type) ∧ (JHipster-App ∨ ¬Client) ∧ (JHipster-App ∨ ¬Server) ∧ (Type ∨
¬JHipster-App) ∧ (Server ∨ ¬JHipster-App) ∧ (Type ∨ ¬Monolithic) ∧ (Type ∨
¬Microservice) ∧ (Type ∨ ¬Gateway) ∧ (Monolithic ∨ Microservice ∨ Gateway ∨
¬Type) ∧ (¬Monolithic ∨ ¬Microservice) ∧ (¬Monolithic ∨ ¬Gateway) ∧
(¬Microservice ∨ ¬Gateway) ∧ (Database ∨ ¬MongoDB) ∧ (Database ∨ ¬SQL) ∧
(Database ∨ ¬Cassandra) ∧ (MongoDB ∨ SQL ∨ Cassandra ∨ ¬Database) ∧
(¬MongoDB ∨ ¬SQL) ∧ (¬MongoDB ∨ ¬Cassandra) ∧ (¬SQL ∨ ¬Cassandra) ∧
(¬HibernateCache ∨ SQL) ∧ (¬Client ∨ Monolithic) ∧ (Client ∨ Microservice ∨
Gateway) ∧ True ∧ ¬False
Conjunctive Normal Form (CNF)
•SAT
•Consistency
•Valid configurations
•Dead features
•…
JHipster v3.6.1: >26k
possible products

•Compact representation of all products
• For instance, using a feature diagram
•Sources for reverse-engineering
• Configurator
• Documentation
• Source code
•Can be formalised for analysis
• SAT of a configuration/product
• Analysis (consistency, dead features, etc.)
• …
Feature Modelling

Yes, most probably
• Hardware
• Plattorm
• Software
• Bundels
• Plugins
• Microservices
• …
• How is it managed (if it is managed)?
• For users
• Documentation only?
• Configurator (e.g., JHipster)?
• For developers
• Model (e.g., feature model)?
• Planned in the development lifecycle?

Variability offers choices.
When we derive a product, we make decisions;
we decide which features will be included in the
product or not. We also say that we bind a decision.

Design-time binding
• Clone and own

Compile-time Binding
• Decided before or at compile time
• Allows optimisations
• Removes unnecessary code from the product
• Once the product has been generated and deployed, it is not
variable any more

Conditional compiling
• Implements variability with preprocessors
• Using annotated code (here, templates)

Conditional compiling
Yeoman
Configuration
Annotated code
Source code 
without annotations

Load-time Binding
• Decided after compilation when the program is started
• More flexibility for reconfiguration
• … but Requires a reboot
• Memory and performance overhead
• All variations are compiled into a single binary
• Consistency conditions must be checked at load time
• Unnecessary functionality may be considered as potential
security threat

Parameters and Configuration Files
• Implement variability using conditional statements (such as if
and switch)
• Controlled by configuration parameters
• Passed to the method or module,
• Set as global variables in a system
Gcc

Run-time Binding
• Decided during program execution and may be changed latter
• Used in dynamic adaptive systems (e.g., robots)
• Most flexible to reconfigure
• Able to react to internal and external stimuli by adapting its behaviour
• Memory and performance overhead
• All variations are compiled into a single binary
• Consistency conditions must be checked at run time
• Unnecessary functionality may be considered as potential
security threat

Design Patterns
• Observer pattern (publish/subscribe) to implement distributed
event handling
• Feature are implemented as observers
• Variability is achieved by registering or not registering observers

Design Patterns
• Template-method pattern
• Using an abstract class
• Implement feature’s behaviour using inheritance

Design Patterns
• Strategy pattern
• Encodes a callback mechanism
• Alternative features where each corresponds to one implementation of the
algorithm
• Uses polymorphism instead of conditional statements

Design Patterns
• Decorator pattern
• Extends object with additional behaviour at run time
• Optional features and feature groups, of which multiple features can be
selected, are implemented as decorators

•Design-time
• Clone and own
•Compile-time
• Conditional compilation
•Load-time
• Program parameters and configuration files
•Run-time
• Design patterns
Variability Binding Time

Yes, most probably
• Hardware
• Platform
• Software
• Bundels
• Plugins
• Microservices
• …
• How is it managed (if it is managed)?
• For users
• Configurator (e.g., JHipster)?
• For developers
• Model (e.g., feature model)?
• Planned in the development lifecycle?

Variability Management section outline
• Examples of variability in your system, and their manifestation in architectural
discussions
• Variability management
• (Partial) feature model
• Current variability implementation mechanism of your system
• Conditional compiling
• Conditional statements (if-then-else)
• Design patterns
• Variability binding time
• Design-time
• Compile-time
• Load-time
• Run-time

• Variability enables adaptability to the business domain
• It should be planned and managed
•Identify development practices
• Clone and own
• Software product line
• Most probably something intermediate
• Binding
• Design, compile, load, and/or runtime binding
• Variability implementation mechanism(s) (conditional compiling,
parameters, design patterns)
•Feature modelling helps to manage variability
• Analysis on the feature model
• Sample products to test
Take-aways

Software Variability Management

More Related Content

Similar to Software Variability Management (20)

More from XavierDevroey (11)

Recently uploaded (20)

Software Variability Management