Computer Science Professor Jo Atlee, Electrical and Computer Engineering Professor Krzysztof Czarnecki, and their former students have been awarded a ten-year most influential paper award to be conferred at VaMoS 2023. Also known as the 17th International Working Conference on Variability Modelling of Software-Intensive Systems, the annual meeting will take place from January 25 to 27, 2023 in Odense, Denmark.
Their paper, A survey of variability modeling in industrial practice, was selected for the prestigious ten-year most influential paper award by the VaMoS 2023 program chairs because of its significant impact, both academically and industrially, on the software engineering community since its publication a decade ago.
L
to
R:
Professors
Jo
Atlee
from
the
Cheriton
School
of
Computer
Science
and
Krzysztof
Czarnecki
from
the
Department
of
Electrical
and
Computer
Engineering
and
cross-appointed
to
the
Cheriton
School
of
Computer
Science.
Jo
Atlee’s
research
is
in
software
engineering
with
a
particular
focus
on
improving
the
quality
of
software.
Her
work
focuses
on
modelling
software
in
terms
of
its
features
and
on
detecting,
resolving
and
managing
feature
interactions,
where
the
goal
is
to
support
the
rapid
development
of
new
features,
services
and
modular
components
that
are
oblivious
of
each
other
yet
interoperate
well.
Krzysztof
Czarnecki
leads
the
Waterloo
Intelligent
Systems
Engineering
Laboratory.
His
current
research
focuses
on
autonomous
driving
and
assuring
the
safety
of
systems
that
rely
on
artificial
intelligence.
He
serves
on
SAE
task
forces
on
level
of
driving
automation,
reference
architecture
for
automated
driving
systems,
verification
and
validation,
and
maneuvers
and
behaviours.
As
part
of
this
research,
he
co-leads
the
development
of
UW
Moose,
Canada’s
first
self-driving
research
vehicle
to
be
tested
on
public
roads
since
2018.
About this award-winning research
Variability modelling — the discipline of representing variability explicitly in dedicated models that describe common and variable characteristics of products in a software product line — is core to many software product line engineering methodologies in academic research and industrial practice. Despite the importance of techniques to model variability little is known about their use, a deficiency that not only undermines their validity but also hinders their improvement.
To address this gap, Professors Atlee and Czarnecki, along with their students and international colleagues, conducted an online survey of industrial variability modelling practices to answer the following questions —
- What variability modelling notations and tools are used?
- What are the scales of industrial models?
- What are perceived benefits and challenges of variability modelling?
The questionnaire also aimed at putting variability modeling into context by identifying the domain and adoption strategies used in product line projects, the artifacts whose variability is described by models, and the roles and experience of respondents. In total, 42 individuals from 16 countries responded, of which 35 were employed in industry, providing an overview of industrial practice.
The survey revealed that industrial product line developers use a much wider application of variability technologies than what was thought to be the case from earlier studies of open-source practice, exceeding simple product configuration and extending to variability management in requirements specification, design, and architecture planning and to product derivation.
The survey also documented a great deal of diversity in the size and extent of detail in variability models. Most respondents used models with fewer than 50 units of variability (i.e., features, variation points, calibration parameters), but about a quarter used models with more than 10,000 units, indicating that variability models have use cases specific to the organization, and that these different requirements need to be satisfied by tools.
The survey also found a great deal of variability in notations and tools, with feature models and tools based on feature modelling clearly dominating. This high diversity also suggested that industry has yet to adopt a gold standard.
Most respondents used re-active and extractive strategies for introducing product lines. In contrast, software product line engineering research primarily focused on pro-active strategies, starting with domain analysis and architectural design. This suggests that the community may need to refocus software product line engineering research towards methods and tools — such as re-engineering and reverse-engineering — that support development of product lines from existing legacy software assets.
Product line engineers undoubtedly face many technical and organizational challenges. The survey revealed that technical challenges primarily include evolution and visualization of models, followed by dependency management. Process-oriented challenges centred on ensuring support in the organization.
To learn more about the research on which this article is based, please see Thorsten Berger, Ralf Rublack, Divya Nair, Joanne M. Atlee, Martin Becker, Krzysztof Czarnecki, Andrzej Wąsowski. A survey of variability modeling in industrial practice. VaMoS ‘13: Proceedings of the Seventh International Workshop on Variability Modelling of Software-intensive Systems, January 2013, article no. 7, pages 1–8. https://doi.org/10.1145/2430502.2430513