Modularizing Software Process Lines

using Model-driven Approaches

A Comparative Study

Fellipe A. Aleixo

1,2

, Uirá Kulesza

, Marília Freire

1,2

, Daniel Alencar

and Edmilson Campos Neto

1,2

Department of Computer Science and Applied Mathematics, Federal University of Rio Grande do Norte, Natal-RN, Brazil

Federal Institute of Education, Science and Technology of Rio Grande do Norte, Campus Natal-Central, Natal-RN, Brazil

Keywords: Software Process, Variability Management, Software Process Lines.

Abstract: This work presents a comparative study of the usage of compositional and annotational approaches in the

modularization of software process lines. In our comparative study, an Open-UP based software process line

extracted from three existing projects are modelled and implemented using the compositional and annotative

approaches with the main aim to address a systematic variability management and automatic process

derivation. The results show that the GenArch-P – annotative approach – can bring many advantages to the

modelling of software process lines considering our comparison criteria.

1 INTRODUCTION

The software process line engineering is inspired in

the software product line engineering (Pohl et al.,

2005) aiming to promote the large-scale reuse of

software processes families. The objective is to

address the automatic customization of software

processes to specific enterprise contexts and

scenarios. Over the last years, several tools and

techniques were developed according with these two

approaches and several empirical studies have

explored and compared their adoption (Kästner &

Apel, 2008) (Kästner, 2010). Much of the evolution

that happened in the software product line

engineering is now being reflected to software

processes domain.

The investigation in software process lines have

grown in the last years and have reached important

results. The consolidation of the theme was achieved

by relevant research work, within such areas: (i)

motivation for software process line engineering

(Rombach, 2005) (Washizaki, 2006) (Ternité, 2009);

(ii) representing process variation (Simidchieva et

al., 2007) (Martínez-Ruiz et al., 2011) (Simmonds &

Bastarrica, 2011); (iii) scoping software process

lines (Armbrust et al., 2009); and (iv)

modularization of software process lines artifacts

(Barreto et al., 2010) (Aleixo et al., 2010). With the

consolidation of this research field, other challenges

emerge and need to be addressed by researchers.

Examples of existing challenges related with the

software process line modularization are: Which

approaches for process modularization are available?

Which techniques and tools should be used in

different scenarios? Which are the benefits and

limitations of each of these approaches?

This work presents a comparative qualitative

study of the usage of compositional and annotational

approaches in the modularization of software

process lines. The compositional approach is

represented by the EPF Composer (EPF, 2012) an

industrial process engineering tool that supports the

modularization and composition of process

elements. The annotational approach is represented

by the GenArch-P (acronym for GenArch-Process)

(Aleixo et al., 2010), an adaptation of an existing

model-based product derivation tool (Cirilo et al.,

2008) that provides support to create generative

models that represents the process variabilities and

associate them with existing process elements.

Our work adapted the comparison criteria from

Kästner (2008) (2010) to analyze these approaches

from different perspectives, such as: modularity,

traceability, error detection, granularity, uniformity

and adoption. Beyond these criteria is included a

new criterion that is the support to systematic

variability management.

The remainder of this paper is organized as

follows. Section 2 provides an overview of the two

120

A. Aleixo F., Kulesza U., Freire M., Alencar D. and Campos Neto E..

Modularizing Software Process Lines using Model-driven Approaches - A Comparative Study.

DOI: 10.5220/0004005801200125

In Proceedings of the 14th International Conference on Enterprise Information Systems (ICEIS-2012), pages 120-125

ISBN: 978-989-8565-11-2

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

investigated approaches. Section 3 describes the

proposed case study, the target software process line

and the used comparison criteria. Section 4 describes

the case study realization and the obtained results.

Finally, the conclusions are presented in Section 5.

2 MODEL-DRIVEN

APPROACHES FOR

SOFTWARE PROCESS LINES

In recent years, several approaches have been

proposed to the development of software process

lines. However there are two of them that stand out

from the others because they focus on the

modularization of process variabilities. One of these

techniques is EPF that allows modularizing process

elements using compositional refinement techniques.

The other one is called GenArch-P that promotes the

variability management of process artifacts using

annotation-based techniques.

2.1 Compositional: EPF Composer

The EPF Composer is a conceptual framework for

authoring, tailoring and deployment of software

processes (EPF, 2012). In EPF, a sofware process is

organized in terms of method content and processes

Method contents describe the process elements, such

as: activities, roles, practices, guidances, among

others. Processes describe the flow of activities,

which reference the elements defined in method

content. Process variabilities in EPF can be handled

with four variability mechanisms that can be

applied to method contents and processes. Those

mechanisms are (EPF, 2012): (i) contributes, (ii)

replace, (iii) extends and (iv) extends and replace.

In contributes variability mechanism, a

specialized element appends the content of a base

element. In replaces variability mechanism, a

specialized element takes place of a base element. In

extends variability mechanism, a specialized element

reuses some attributes, and overwrite others of the

base element. At last, the extends and replace

variability mechanism combines the effects of the

extends and the replaces variability mechanisms into

one strategy. While the replace variability replaces

all contents from the base element, this variability

mechanism allow the inheriante of the non-defined

attributes of the specialized element.

2.2 Annotation: GenArch-P

GenArch-P (Aleixo et al., 2010) is a tool for

software process derivation adapted from its original

version to software product lines, called GenArch

(Cirilo et al., 2008). The tool is based on the

annotational approach, where the variable elements

of a software process are annotated to reflect which

type of variation is implement by the element. The

annotational approach is widely used in various

software product lines tools, such as pure::variants

and CIDE.

The GenArch-P works with two models: a

simplified process model and the feature model. The

elements of the process model are annotated to

reflect their relationship to specific features. The

annotations in GenArch-P are based on the

underlying structure of the process model where

each element can have an associated variation

property that describes the variation type of this

element and the feature parent.

The process model is a simplified representation

of the software process specification, that presents

the process elements as a hierarchical tree. The

feature model is automatically generated by the tool.

After the annotations, the process model acts like a

configuration model, allowing the visualization of

the associations between features and process

elements. The feature model is a common artifact at

the software product lines development, introduced

by Feature-Oriented Domain Analysis (Kang et al.,

1990), that represents all the possible instances of a

product line. The model presents all the possible

options to be selected, in terms of features.

3 STUDY SETTINGS

In our study, we have modelled the same OpenUP-

based process line composed of different kinds of

process variabilities using the evaluated modelling

approaches. After the specification of the software

process line using EPF and GenArch-P, we have

conducted a comparative analysis of the final results

based on criteria adapted from a previous study

(Kästner et al., 2008) (Kästner, 2010).

3.1 Target Software Process Line

The definition of the software process line used in

our study involved the analysis of OpenUP-based

processes from three existing research and

development projects. These projects were

developed in cooperation between our institution

and other ones. The first project was the

development of a software system for auditing the

telephony networks. This system performs the

ModularizingSoftwareProcessLinesusingModel-drivenApproaches-AComparativeStudy

121

counting, summarization, and the analyses of the

connection records created by a specific hardware.

The second project involved the development of a

module of a distributed system responsible for the

collection and the storage of the information related

to the federal institutions of professional and

technological education in Brazil. The third, and

last, project comprised the implementation of an

integrated academic and administrative management

system for the federal institutions of professional

and technological education in Brazil.

The complete specification of the software

process line to be modularized was accomplished

using the extractive technique. It comprised the

analysis of the commonalities and variabilities of the

selected software processes. The common process

elements compose the core of the process line, while

the variable process elements were modelled

separately and associated to high-level features. It

was identified 76 process elements composing the

core of the process line. Regarding the variabilities,

it was found 9 optional features, 8 alternative

features, and 5 OR-features. Table 1 shows the

names of the identified features and the number of

associated process elements.

Table 1: Identified features in the software process family.

Feature

Alternatives

(if there is)

Process

elements

Others processes influences

Additional elements from the Scrum Process 25

Requirements techniques and technologies

Specification

techniques

Use cases 27

Users stories 27

Product backlog 27

Specification

tools

Asta Community 24

Rational Software Architect 24

Borland Together 24

ArgoUML 24

Design techniques and technologies

Architecture

documentation

Agile design 64

Well documented architecture 16

Implementation techniques and technologies

Used language

Java 7

C# 7

Ruby 7

Phyton 7

Use of JEE framework 4

Use of Eclipse IDE 5

Use of JUnit framework 15

Continuous integration techniques and technologies

Use of Hudson tool 8

Metrics techniques and technologies

Use of Maven tool (code metrics – SVN mined) 5

Metric for assess the activities progress 3

Metric for asses the deadlines fulfilment 3

Metric for asses the duration of main activities 3

During our analysis, we have also found

constraints between features from the process line.

For example, the features representing the usage of

JEE and JUnit frameworks requires the selection of

Java features as the programming language. Further

information about the specification of the software

process line using the different approaches can be

found here (Aleixo et al., 2012).

3.2 Comparison Criteria

In order to promote the assessment of software

product line implementation techniques, Kästner et

al. (2010) have defined comparison criteria with

such purpose. In our work, we have adapted most of

these criteria – modularity, traceability, error

detection, granularity, uniformity, and adoption – to

the context of software process lines, which are

presented in Table 2. In addition, we have also

analysed the approaches support for systematic

variability management.

Table 2: Comparison criteria.

Criterion Definition

Modularity

It analyses the support to modularization of the

process elements associated with specific

features (features implementations), isolating

the implementation of a specific feature

Traceability

It shows how easy is the visualization of the

mapping between features and their related

process elements

Error detection

It analyses how existing approaches provide

support to consistency checking of the

software process line and their (resultant)

derived processes

Granularity

It assesses the approach support to associate

features with process elements of different

granularity, considering also the attributes of

the process elements

Uniformity

It analyses the ability to uniform support for

different forms of software processes

specification, evaluating how an approach is

tied to a specific process specification form

Adoption

It determines the difficulty of adopting existing

approaches in terms of the amount of

necessary knowledge (concepts, mechanisms

and tools) for the approach application

Systematic

variability

management

It assesses the approach mechanisms for

systematic and effective management of

variability: (i) the variability specification,

their constraints, and the mapping with process

elements; and (ii) the approach support to

automated process derivation from existing

process core assets

4 STUDY RESULTS

In this section, we report our study results by

describing the process line modelling using EPF

Composer (Section 4.1) and GenArch-P (Section

4.2). In addition, we also present and discuss the

obtained results for the comparison criteria of the

two approaches (Section 4.3).

ICEIS2012-14thInternationalConferenceonEnterpriseInformationSystems

122

4.1 Process Line Modelling using EPF

Composer

Process Line Engineering. The modelling of the

process line was accomplished taken as basis the

original method plugin of OpenUP. EPF defines a

customizable software process in terms of method

contents and processes, which are fundamental

concepts from EPF. The method contents allow

specifying the process elements and all their

attributes, which can later be (re)used and composed

to define the workflows of a new process.

The process line engineering involved the

following steps: (i) creation of a new method plugin;

(ii) creation of the mandatory process elements, in a

“core” content package; (iii) creation of specific

content packages to each feature, with the

correspondent process elements; (iv) creation, if

necessary, of the correspondent capability patterns to

each feature, to encapsulate the additions to the

workflow defined by the “core” elements. Each of

these capability patterns uses the content variability

"contributes" mechanism to refine the capability

pattern that represents the core including the specific

flow associated with the process variability. Table 3

presents a summary of the EPF mechanisms used to

implement each type of feature.

Table 3: EPF mechanisms used to each type of feature.

Type of

feature

Used EPF mechanism

Alternative

(i) hierarchical content package structure +

(ii) process elements of each alternative +

(iii) capability patterns to encapsulate the

additions to the “core” workflow

Optional

(i) content package structure +

(ii) associated process elements +

(iii) capability patterns to encapsulate the

additions to the “core” workflow

OR-feature

(i) hierarchical content package structure +

(ii) process elements of each related option +

(iii) capability patterns to encapsulate the

additions to the “core” workflow

Process Derivation. EPF Composer provides the

functionality to the definition of a process

configuration. In the configuration definition, the

process engineer chooses which modular structures

will be part of a published process – a navigable web

site. The process of configuration definition and

process instance publication can be repeated,

allowing the publication of all process line

possibilities. After the configuration definition, a

customized process is published with its respective

workflows and process elements (activities, tasks,

roles, among others).

4.2 Process Line Modeling using

GenArch-P

Process Line Engineering. The process line

engineering involved the following steps: (i)

definition of a complete software process

specification (ii) creation of a new GenArch-P

project; (iii) use the tool to parse the process

specification, generating a simplified process model;

(iv) annotation of the involved process elements

with features expressions defining the configuration

knowledge; (v) specification of constrains and

dependency relationship between features.

Figure 1 shows an example of the GenArch-P

process model where the elements are hierarchically

distributed to compose a wide view of the software

process structure. Figure 1 also shows the

annotations in process elements defining

associations to the variabilities of the process line. It

represents the configuration knowledge of mapping

from features to process elements. For example, the

feature that represents the agile design alternative

feature is associated to the following process

elements: (i) concurrent testing, (ii) continuous

integration, (iii) test-driven development, and others

not presented in the figure. After the annotation of

process elements, the GenArch-P generates the

correspondent feature model. Both the feature model

and the annotated process model guide the tool in

the automated process instance derivation.

Figure 1: Fragment of the process model with the features

annotations.

Process Derivation in GenArch-P starts with the

creation of a new feature configuration that allows

the processes engineers to select the desired features

for a new process instance. During the features

selection, constraints associated with features are

analysed. Thus, the selection of specific features

ModularizingSoftwareProcessLinesusingModel-drivenApproaches-AComparativeStudy

123

may imply the removal of others. After the features

selection, a new process specification is derived,

with the process elements associated to the chosen

features.

4.3 Criteria Analysis

Modularity. Using EPF Composer is possible to

group software process elements in specific content

packages, and workflows in capability patterns.

These modularization mechanisms allow the

grouping of process elements related to a given

feature. Thus, our study concludes that EPF

Composer provides useful support for modularity of

process specifications. GenArch-P does not provide

modularization mechanisms for the process

specifications. The process engineers only interact

with a simplified model of the process, abstracting

the details and organization of the process

specification. We concluded that GenArch-P has

partial support for modularization, because the

software process specification could already be

modularized with low-level mechanisms.

Traceability. Using the EPF Composer the

traceability is achieved by the organization of the

elements in the correspondent structures: content

packages and capability patterns. Although the EPF

Composer lacks an explicit mechanism to map

features to process elements, we can conclude that

EPF Composer has a partial support for traceability.

Using GenArch-P, the process engineer can

visualize all mapping relationships between features

and process elements in the annotated process

model. Due to this characteristic, it is concluded that

GenArch-P offers good support for traceability.

Error Detection. The EPF Composer does not

offer a mechanism to detect semantic errors in the

specification of the process line. In addition, EPF

cannot represent constraints between features. The

only initiative accordingly is a dependency checking

during the configuration definition. Because of that,

we can conclude that the EPF Composer offers a

weak support for error detection. GenArch-P does

not have an explicit mechanism for error detection,

but during the annotation of the process elements, it

can specify constraints associated with the features.

These constraints allow guaranteeing that existing

constraints between features will be respected, such

as requires or excludes relationships. Thus, we can

say that GenArch-P provides currently only partial

support for error detection.

Granularity. The content package and capability

pattern mechanisms from EPF Composer provide

support to coarse-grained granularity through the

grouping of process elements and workflows.

Moreover, the variability mechanisms can be used to

redefine existing attributes of process elements, thus

providing support to fine-grained granularity. We

conclude that EPF Composer has a good support for

coarse and fine-grained elements. GenArch-P

restricts the granularity of the process elements at

the level of elements captured by the parsing of the

process specification in order to generate the

simplified process model. The current version of

GenArch-P only supports the granularity of process

activities and tasks, but not yet their attributes. We

concluded that GenArch-P currently provides a

partial support for fine-grained granularity.

Uniformity. EPF Composer does not provide

uniform support for different software process

specifications. It happens because the EPF

Composer only specifies software processes

according with the UMA meta-model. Because of

that, it was concluded that the EPF Composer does

not offer uniform support for different forms of

software process specification. Using GenArch-P,

the process specification language is abstracted by a

simplified process model. This model has not any

dependency with the low-level process specification.

Because of this, it is concluded that the GenArch-P

provides support to the uniformity criterion.

Adoption. The EPF Composer provides an

ample set of concepts and mechanisms that need to

be known by the process engineers. In addition, they

also need to understand the variability mechanisms

to address the modularization and configuration of

process variabilities. By forcing a process engineer

to know a set of mechanisms, concepts and

functionalities, it is concluded that EPF Composer

has weak support for the adoption criterion. The

GenArch-P tool automates most of the tasks

involved in the modelling of the software process

line. It does not require that the process engineers

have an extensive knowledge beyond the software

process variabilities, allowing an easy adoption.

Systematic Variability Management. Using

EPF Composer, the process engineer cannot specify

the existing variabilities using a feature model and

explicitly associate them to process elements. On the

other hand, GenArch-P allows an explicit feature

modelling, including the feature representation with

its respective constraints, the mapping to existing

process elements, and the automatic process

derivation based on feature selection. Thus, we can

conclude that GenArch-P offers a systematic

variability management.

ICEIS2012-14thInternationalConferenceonEnterpriseInformationSystems

124

5 CONCLUSIONS

This paper presented the results of a comparative

study of compositional and annotational modeling

approaches of software process lines. Two modern

approaches were selected: (i) EPF Composer –

representing the compositional approach and (ii)

GenArch-P – representing the annotational

approach. These two investigated approches were

used to specify a non-trivial Open-UP processes

line.

Our study adopted a comparison criteria

previously adopted in the analysis of the

implementation techniques of software product lines

(Kästner, 2010). Based on the results of the study, it

can be concluded that the annotational approach

obtained better results in the software processes lines

definition. In five of the seven defined criteria, the

GenArch-P presented better results, which are: (i)

traceability, (ii) error detection, (iii) uniformity, (iv)

adoption, and (v) systematic variability

management. The EPF Composer had better results

in the modularity criterion, which reinforces one of

the known strengths of compositional approaches. In

the granularity criterion, the EPF Composer

approach had also better results, due to the variety of

variability mechanisms provided.

The study illustrated that annotative and

compositional approaches have their own strengths

and limitations defining software process lines, and

both are valid alternatives. The possible integration

of the compositional and annotative approaches can

combine the strengths of these two approaches and

will be investigated in future work.

ACKNOWLEDGEMENTS

This work was partially supported by the National

Institute of Science and Technology for Software

Engineering (INES) - CNPq under grants

573964/2008-4 and CNPQ 560256/2010-8.

REFERENCES

Aleixo, F. A., Freire, M. A. & Kulesza, U., 2012. Software

Process Lines. [Online] Available at: https://sites.

google.com/site/softwareprocesslines/ [Accessed 27

January 2012].

Aleixo, F. A., Freire, M. A., Santos, W. C. d. & Kulesza,

U., 2010. A Model-driven Approach to Managing and

Customizing Software Process Variabilities. In 12th

ICEIS. Funchal, Madeira, Portugal, 2010. SciTePress.

Aleixo, F. A., Freire, M. A., Santos, W. C. d. & Kulesza,

U., 2010. Automating the Variability Management,

Customization and Deployment of Software

Processes: A Model-Driven Approach. Lecture Notes

in Business Information Processing, pp.372-87.

Armbrust, O. et al., 2009. Scoping software process lines.

Software Process: Improvement and Practice, 14-3,

pp.181-97.

Barreto, A., Duarte, E., Rocha, A. R. & Murta, L., 2010.

Supporting the Definition of Software Processes at

Consulting Organizations via Software Process Lines.

In 7th QUATIC. Porto, Portugal, 2010. IEEE

Computer Society.

Cirilo, E., Kulesza, U. & Lucena, C. J. P. d., 2008. A

Product Derivation Tool Based on Model-Driven

Techniques and Annotations. The Journal of Universal

Computer Science, 14-8, pp.1344-67.

EPF, 2012. Eclipse Process Framework Project (EPF).

[Online] Available at: http://www.eclipse.org/epf/

[Accessed 27 January 2012].

Kang, K. C. et al., 1990. Feature-oriented domain analysis

(FODA) feasibility study. SEI.

Kästner, C., 2010. Virtual Separation of Concerns:

Toward Preprocessors 2.0. Magdeburg, Germany:

Dissertation, Otto-von-Guericke-Universität.

Kästner, C. & Apel, S., 2008. Integrating Compositional

and Annotative Approaches for Product Line

Engineering. In GPCE Workshop on Modularization,

Composition and Generative Techniques for Product

Line Engineering (McGPLE). Passau, Germany, 2008.

University of Passau.

Kästner, C., Apel, S. & Kuhlemann, M., 2008. Granularity

in software product lines. In ICSE., 2008.

Martínez-Ruiz, T., García, F., Piattini, M. & Münch, J.,

2011. Modelling Software Process Variability: an

Empirical Study. IET Software, 5 (2), pp.172-87.

Pohl, K., Böckle, G. & Linden, F. v. d., 2005. Software

product line engineering: foundations, principles, and

techniques. Berlin, Germany: Springer-Verlang.

Rombach, H. D., 2005. Integrated Software Process and

Product Lines. In ISPW. Beijing, China, 2005.

Springer.

Simidchieva, B. I., Clarke, L. A. & Osterweil, L. J., 2007.

Representing Process Variation with a Process Family.

In ICSP. Minneapolis, MN, USA, 2007. Springer.

Simmonds, J. & Bastarrica, M. C., 2011. Modeling

Variability in Software Process Lines. Santiago, Chile:

Universidad de Chile.

Ternité, T., 2009. Process Lines: A Product Line

Approach Designed for Process Model Development.

In 35th Euromicro Conference on Software

Engineering and Advanced Applications. Patras,

Greece, 2009. IEEE Computer Society.

Washizaki, H., 2006. Building Software Process Line

Architectures from Bottom Up. In PROFES.

Amsterdam, The Netherlands, 2006. Springer.

ModularizingSoftwareProcessLinesusingModel-drivenApproaches-AComparativeStudy

125