Software Reuse

1
Software Reuse

• Cornelia Boldyreff
• Department of Computing and Informatics

2
Overview

• Software Reuse
• Metrics
• Operationalizing Reusability Guidelines
• More Flexible Support for Reuse in OSS
• Future Directions

3
Customer Requirements
RA
Spec/n
Maintenance
BIN
Design
Testing
Impl/n
4
Customer Requirements
RA
Spec/n
Maintenance
Design
Testing
Impl/n
5
Two possible definitions of Software Reusability

• 1 - the extent to which a software component can
  be used (with or without adaptation)
• in multiple problem solutions
• 2 - the extent to which a software component can
  be used (with or without adaptation)
• in a problem solution other than the one for
  which it was originally developed.
• where a software component refers to any part of
  a software lifecycle product, e.g. Z schema, data
  flow diagram, a code module, etc.

6

• Definition 1
• - suggests planned effort
• - is more productivity oriented
• - recognises the need to emphasise reuse as meta
  activity in development
• Definition 2
• - implies reuse is accidental and ad hoc
• - is a more literal definition
• - characterises past achievements
• may lead people to think we already do that
• My view is that software reuse is simply a
  metaphor for intellectual progress in software
  engineering.

7
Examples of Direct Reuse of Software Components

• Published algorithms
• OO Design Patterns from the Gang of Four
• Architectural Frameworks
• Standard programming languages
• Libraries of programs, modules, functions,
  subroutines
• Standard packages
• COTS
• Web Services
• This is also simply called component reuse.

8
SPSS
grep
fork()
LDX 4 BRN J1
if xgt0 then ...
Sqrt()
1001001
Binary code
Assembler Source Code Fns.
Libraries Programs Packages
levels of abstraction
Each level of abstraction supports a higher level
of reuse.
9
Indirect Reuse of Software(also called
Generative Reuse)
This can involve the use of books on software
design principles and design theory as well as
case studies of specific software. E.g. Books on
compiler construction This can also involve the
use of supporting software tools which generate
software, e.g. lex - lexical analyser generator
and yacc - compiler generator (which generates a
syntax analyser and associated actions).
Language description Compiler Generator
Compiler
10

• Most software engineers implicitly reuse existing
  software components and concepts - but how can we
  make reuse within the development process more
  explicit?

The Reuse Process
Software Reuse Support/Admin
Development with Reuse
Development for Reuse
Building up the reuse resources is a necessary
1st step, only then can reuse become an explicit
part of the SLC.
11
Why reuse software?

• Quality - we can produce better software
• Productivity - we can produce software faster
• Progress - we can tackle bigger problems by
  building on known solutions
• So how can we measure reusability??

12
Software Metrics

• SE is the application of a systematic,
  disciplined, quantifiable approach to the
  development, operation and maintenance of
  software (SCC, Gibbs)
• Early metrics focused on software complexity
  (McCabe, Halstead) - product metric and software
  cost estimation (Boehm) - process metric

• The advent of the OO paradigm has led to specific
  OO metrics, both generic and language-specific.
• Developments have led to the use of metrics in
  software quality assurance (the -abilities) and
  to inject more tractability and visibility into
  the SLC, thus improving control and management.

13
Operationalising Reusability Guidelines

• Analysis existing Ada Reusability Guidelines and
  categorise under 4 properties Adaptability,
  Comprehensibility, Independence and Robustness.
• Identify corresponding metrics for each property
  (G-Q-M) and weighting strategy.
• Validate through application in a case study of
  Ada components.

14
Research to apply this in the OSS context

• Mining OSS projects to examine their directory
  structures and dependences between source code
  files in the directories
• Hypothesis is that such analysis will enable us
  to detect actual and potentially reusable OSS
  components
• Why directories? In the absence of designs, we
  assume these reflect the hierarchical
  decomposition of the system.

15
Future Directions

• Building support for reuse in the OSS context is
  difficult no central repository, nor management
  and support.
• Even company based reuse programmes can suffer
  from lack of support and maintenance.
• Why not try for a google like service based on
  consolidated analysis of large OSS repositories,
  e.g. SourceForge, Savannah, FreshMeat, etc?

16
Reuse and Evolution

• Learning from the Past
• Whats already been done
• What we know/How we did it
• Laying the foundations
• Codification of experience
• Agreement on accepted practice
• Abstraction of theory

• Payback and Feedback in the Future
• Repeating what we did before more efficiently
• Tackling new situations
• Refining and extending whats known

17
Background

• Software Engineering is a discipline concerned
  with the development of software systems.
• Its goal is continuous improvement of
• software technologies and products,
• software development processes and methods, and
• Software Engineering as a profession.
• It aims effectively to improve both software
  development theory and practice.

18
Key themes

• The evolution of Software Engineering
• its origin in the Software Crisis
• its emergence as an academic discipline
• its role as an enabler of society as we know it
• its future prospects.
• My underlying research in Software Reuse,
  Maintenance, and Software Evolution and work on
  Software Engineering Environments

19
The Software Crisis

• Over 35 years ago, in Autumn 1968, the Software
  Crisis was acknowledged, and the establishment
  of Software Engineering was proposed as a means
  of solving the Software Crisis.
• Although software engineers can now solve the
  problems that gave rise to the original crisis,
  the software industry suffers chronically from
  software crises as it develops new ways of
  working, new technologies, and innovative
  software applications of increasing complexity
  and scale.
• In some consumer products, the software component
  is doubling yearly. The web-enabled fridge is
  here!

20
(No Transcript)
21
Current Situation

• Programming has been an established profession
  undergoing refinement for more than 50 years. The
  British Computer Society, the UK professional
  organisation for Computing, celebrates its 50th
  anniversary this coming year.
• Academics have made progress in establishing
  Software Engineering as an academic discipline
  and in instituting software engineering degrees.
• Industrially led research has made progress
  towards improved models, methods, and tools for
  software engineering.

22
Inaugural Meeting of the BCS Lincolnshire
Sub-branch

• 30 May EEMTEC Building 630 p.m. for 7 p.m.
• Debate Debate This house holds that Computing
  is currently at best a semi-profession.
• This is taken by adapting Beynon-Davis'
  observation that "information systems is
  currently at best a semi-profession", The
  Computer Bulletin, September 1999, as quoted in
  Frank Bott's book Professional Issues in
  InformationTechnology.
• Speakers Philip Levine, MBCS, CITP (Proposer)
  and Adrian Wheal, Ceng, FBCS, CITP (Opposer)

23
Software Engineering Textbooks

• Kernighan and Plauger's Software Tools teaches
  how to write good programs that make good tools,
  i.e. programs which have proved valuable in the
  production of other programs (1976)
• Sommerville's Software Engineering (1982),
  currently in its 7th edition
• McDermid's Software Engineer's Reference Book
  (1991)

24
Software Engineering is Applied Computer Science

• Engineering is a human endeavour and thus it is
  subject to error. Petroski, TO ENGINEER IS
  HUMAN The Role of Failure in Successful Design,
  1982
• There is no silver bullet (Brookes)!
• Software Engineering has an empirical basis - we
  always need applied research using industry as
  laboratory (Potts)!

25
Software in Society

• There is hardly any aspect of human life that is
  not supported in some way by software systems
  the Internet, telecommunications, commerce,
  government, entertainment, transportation,
  education, health care, the home and increasingly
  the individual.
• As our dependence of software grows and our
  software systems become more interconnected, the
  challenges of developing, i.e. modelling,
  maintaining and evolving, these systems become
  greater.

26
New Approaches

• Community based Open Source Software development
• Increased globalisation of the software industry
• More agile approaches encouraging wider user
  participation in the software development
• New forms of distributed development

27
Future Prospects

• Self-adapting software systems are becoming a
  reality, e.g. self-healing networks.
• The Internet and its development into the Grid
  provides us a world computer. This is an
  expanding network of all computers and
  computational devices interconnected.
• Intelligent cities, buildings, homes, wearable
  computers are now becoming more common leading to
  pervasive systems.
• Self-assembling dynamically reconfigurable
  systems where the components, typically service
  based, are made self-aware through software.
• Trade-offs between moving data or computation
  through networks

28
Caida/Science Photo Library Internet Traffic
29
Software Complexity

• One program one computer software failures are
  discrete events and these are essentially closed
  systems.
• N programs N computers each program breaks
  down in different ways depending in part on
  conditions under which it has been run.
• Large distributed software systems consist of
  many interconnected programs and data sources
  running over networks of computers. These systems
  are essentially open systems. They dynamically
  change as new services and computer are added or
  removed.
• Modelling and managing such complexity is a key
  challenge!

30
Software Reuse

• Software reuse has been inspired by Alexander's
  work on design patterns in architecture.
• By studying existing software systems within a
  particular domain, common design principles and
  software design concepts can be abstracted for
  reuse in the development of future systems.
• Software reuse is a metaphor for intellectual
  progress in Software Engineering.

31
(No Transcript)
32
Source UK Steel Association
33
(No Transcript)
34
(No Transcript)
35
Software Evolution

• Software is created out of nothing but pure
  thought but software developers draw on the huge
  wealth of existing software principles, possibly
  embodied in tools, and concepts.
• Every software development starts with a blank
  document and its subsequent maintenance.
• Once created evolutionary software has the
  potential to change in its environment and it
  must be evolved in response to changes within its
  environment if it is to continue to be useful.
• Component/Service based systems exhibit
  multi-dimensional evolution. Supporting change
  awareness in such systems as they undergo
  development is an exciting research challenge!

36
The closed Loop Concept of Systems Engineering
The closed Loop Concept of System Engineering
From Alexander W. Boldyreff, 1954
37
ECMA Reference Model for Software Engineering
Environments (ECMA 1993) (taken from
Sommervilles slides)
38
Work on SEEs

• IST IPSE, ISTAR, mid 80s (Unix based)
• ESPRIT Practitioner Project, late 80s (Unix)
• ESPRIT AMES Project, mid 90s (early Web)
• EPSRC CARD Project, late 90s (closed n/w)
• Framework 5 GENESIS Project and EPSRC CoDEEDS
  project, 2000s (Web service based OSS)
• Current research on Eclipse based awareness
  support (OSS)
• Base Case Models, Methods, Tools and Evaluation
  through usage
• Inductive Step Theoretical Improvement based on
  feedback and advances in Technology allows
  progression from n to n1

39
SEE of the future?!
40
Thank you

• Dedicated teachers, inspiring and brilliant
  colleagues, exceptional students
• Supportive family and friends and
• Members of my research group, past and present.

41
Questions