Distributed Software Engineering Research Group, Department of Computing and Informatics, University

1
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Research Challenges in Open Source Software
Cornelia Boldyreff Distrubuted Software
Engineering Research Group Department of
Computing and Informatics University of Lincoln
Talk Outline
Open Source Software What, How, and
Why Research within CALIBRE Collaborativ
e Work Environments from IPSE to CWE Other
FLOSS related research at Lincoln
2
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
What is Free Libre Open Source Software?
Background

• Early UNIX based developments and associated
  software tools
• MIT AI Lab, Stallman's GNU project in 1982 and
  the Free Software Foundation
• Linus Torvalds' Linux circa early 1990s
• Key feature of Linux built by volunteers
  co-ordinating over the internet
• By 1993, a stable and reliable Linux attracted
  ports of commercial applications

Key features of FLOSS
The free (as in libre) aspect of OSS provides for
a more democratic participation in software
development by a wider community. As the source
os software is freely available is intellectually
accessible as well as physically accessible. OSS
can also be free (as in gratis) by being provided
at no (or low) cost, enabling anyone to obtain
the software for little cost. This especially
attractive to those in developing countries.
3
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
How is FLOSS developed?
Variety of OSS projects

• The large and successful projects Linux,
  Appache, Mozilla, Open Office, Eclipse
• Large distributed development projects with
  their own release strategies and practices
• Range of projects
• Systems Software OS, DBMS, Web servers/browsers,
  Languages,
• Applications needed by all editors, office
  software, e-learning applications
• and contents, e-government, health informatics

Various Surveys and Studies
The EU FLOSS Surveys Software engineering
researchers mining FLOSS project repositories for
data to study software evolution in OSS and also
quality of OSS products and processes more
generally, e.g. Recent ICSE workshop on this
topic. In particular, this research aims to
facilitate the use of agile practices within
open source software development.
4
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Motivation Pushing The Boundaries of Agility
The Troubles With Agile Practice

• The Agile methods are just a strict as their
  classic counterparts
• The usefulness of Agile practice is limited
• We can infer a requirement of team collocation
• Agile practice is best suited to small
  development teams

The Aim Of This Research
The main aim of this research is to extend the
number of useful applications of agile practice
through the loosening of the collocation and
small development team requirements. In
particular, this research aims to facilitate the
use of agile practices within open source
software development.
5
(No Transcript)
6
CALIBRE Key Goals

• To integrate and coordinate libre software
  research and practice to ensure that the libre
  phenomenon flourishes and delivers to its true
  potential, especially for the European
  secondary software sector
• To foster the effective transfer of the many
  useful lessons from libre software to facilitate
  the next generation of software engineering
  methods and tools
• To establish a European industry open source
  software research policy forum CALIBRATION

7
Background to CALIBRE

• Libre Software, Agile Methods Distributed
  Development key areas for future of software
• Libre software phenomenon very strong in Europe
  but not determining strategic policy
• Most emphasis has been in government public
  sector (e.g. COSPA project) rather than secondary
  software sector
• Libre software success not guaranteed paradoxes
  abound!
• Collectivist v. individualist cult v.
  establishment talented developers for free
  proprietary co.s profiting most etc
• Hype overselling
• Complex phenomena need reasoned intervention from
  multi-disciplinary perspective

8
CALIBRE Consortium Work Packages

• Partners
• UL University of Limerick
• URJC Univ Rey Juan Carlos
• UM Univ of Maastricht
• UPMC University Pierre-et-Marie-Curie
• BICST Business Innovation Centre of Alto
  Adige-Südtirol
• UCC Univ College Cork
• LIN Univ of Lincoln
• GET Groupe des Ecoles de Telecommunications
  
• PUT Poznan Univ of Technology
• MAC National Microelectronics Applications Centre
  Ltd
• CHI Chinasoft
• SKO University of Skövde

• Work Packages
• WP1 Characterizing Libre Projects, Products
  Processes
• WP2 Distributed Software Development
• WP3 Agile Methods
• WP4 Co-ordination, Collaboration, Education
  Dissemination
• WP5 Project Management

9
Fundamental Challenges

• Moving beyond evangelical anecdotal
• Building sustaining libre software communities
• Effecting European strategic planning for libre
  software
• Understanding new libre business models modes
  of organising
• Third way - build, buy or libre
• Hybrid mix of libre proprietary software
• Legal, IPR and licensing issues
• Symbiotic collaboration co-opetition
• Stimulating libre development in vertical domains
• Updating software development paradigm theory
  practice
• Overall impact of CALIBRE to address these
  challenges

10

• Workshops/Conference
• Case studies of success
• Action Research studies cycle 1
• Database of libre data
• New business models
• Quality of libre
• CALIBRATION Industry Policy Forum established
• Lessons to/from libre to distrib dev agile
• CALIBRE Knowledge Base

• Workshops/Conference
• Action Research studies cycle 2
• Proven libre business models
• Libre Euro-focused strategy formulation
• Stimulation of libre in new domains
• Educational programmes
• Roadmap for future research in libre, distrib dev
  agile methods

Year 1
Year 2
11
Anticipated CALIBRE Impacts

• Its easier to create the future than predict
  it! Alan Kay
• Moving from evangelical/anecdotal to coherent
  scientific research roadmap agenda
• Putting libre on industry agenda (esp secondary
  software sector) with industry-research policy
  forum
• Best practice business models
• Integrating libre, agile distributed
  development approaches into new software
  development paradigm
• More effective and coherent practice-informed
  research
• Stronger sustainable libre software communities
  in new domains
• Leading to
• more competitive European industry
• more coherent European research

12
Anticipated Impacts

• Its easier to create the future than predict
  it!
  - Alan Kay
• Moving from evangelical/anecdotal to coherent
  scientific research roadmap agenda
• Putting OSS on industry agenda (esp secondary
  software sector) with industry-research policy
  forum
• Best practice business models
• Stronger sustainable OSS software communities in
  new domains
• More effective and coherent practice-informed
  research
• Integrating OSS, agile distrib dev approaches
  into new s/w development paradigm

12/8
13
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Motivation Pushing The Boundaries of Agility
The Troubles With Agile Practice

• The Agile methods are just a strict as their
  classic counterparts
• The usefulness of Agile practice is limited
• We can infer a requirement of team collocation
• Agile practice is best suited to small
  development teams

The Aim Of This Research
The main aim of this research is to extend the
number of useful applications of agile practice
through the loosening of the collocation and
small development team requirements. In
particular, this research aims to facilitate the
use of agile practices within open source
software development.
14
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Motivation Pushing The Boundaries of Agility
The Troubles With Agile Practice

• The Agile methods are just a strict as their
  classic counterparts
• The usefulness of Agile practice is limited
• We can infer a requirement of team collocation
• Agile practice is best suited to small
  development teams

The Aim Of This Research
The main aim of this research is to extend the
number of useful applications of agile practice
through the loosening of the collocation and
small development team requirements. In
particular, this research aims to facilitate the
use of agile practices within open source
software development.
15
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Motivation Pushing The Boundaries of Agility
The Troubles With Agile Practice

• The Agile methods are just a strict as their
  classic counterparts
• The usefulness of Agile practice is limited
• We can infer a requirement of team collocation
• Agile practice is best suited to small
  development teams

The Aim Of This Research
The main aim of this research is to extend the
number of useful applications of agile practice
through the loosening of the collocation and
small development team requirements. In
particular, this research aims to facilitate the
use of agile practices within open source
software development.
16
Research Scope

• The current trend towards distributed teams in
  software engineering
• leads to several problems
• Coordination problems due to lack of informal
  communication
• Processes do not often address informal
  communication needs, concentrating on exchange of
  deliverables/milestones etc
• Developers improvise with e-mail/instant
  messaging etc
• CSCW has met with mixed success, sometimes even
  hindering the process of collaboration in
  some domains (Lurey et al)
• We suspect that this is an integration issue
  most general purpose CSCW tools are flexible but
  require significant effort on the part of the
  user to provide useful, informal communication.
• e.g. instant messaging provides presence
  awareness, but no gaze awareness. User
  must explicitly inform others of their actions

17
Awareness

• One type of informal communication lost in a
  distributed context is
• awareness, of others' presence and their actions
  both current and past.
• In a colocated context, awareness is easy to
  establish merely watch one's colleagues
• Various CSCW tools have attempted to recreate
  aspects of awareness, creating types such
  as presence, gaze and workspace awareness etc.
• These are usually synchronous (real-time) or
  asynchronous
• Few present any contextual information about a
  group of artefacts.
• Often the user must expend effort to find
  pertinent information.
• Therefore we have defined historical awareness
  which attempts to
• combine the advantages of existing awareness
  types
• "The complete context of an artefact's creation,
  derived from a collection of heterogeneous
  artefacts (source code, design etc) rather than a
  contextless view of a single artefact's
  evolution."

18
P2P with semantic overlay for information exchange

• Reflector based CSCW systems are not resilient
  (SPOF) and not scalable.
• Exchange of awareness information may benefit
  from a more robust
  communication architecture such as P2P.
• With the addition of a semantic overlay network,
  the requirements for awareness
  pertinence scalability may be addressed
• The nearest peers to any client will contain the
  most relevant information less network
  load
• A new algorithm for network self-organisation has
  been developed. This does not rely on
  "Super Peers" like other semantic overlay
  implementation.
• This keeps setup complexity down.
• The exchange of awareness information over this
  P2P architecture has been tested
  successfully in simulation.
• Likewise the network performance in poor
  conditions (links breaking etc).

19
Eclipse Integration

• Now that the awareness exchange has been proven
  in simulation, a
• prototype tool must be implemented and evaluated.
  
• The proposed tool will be integrated with the
  eclipse environment
• Ease of data collection
• Existing, familiar, software engineering tool
  with large user community
• Highly extensible already.
• Each participating Eclipse workbench will be a
  node in a P2P network using the semantic
  overlay described above
• The tool will generate and transmit awareness
  information based on what the user is doing,
  including
• Open resources in the workbench
• Content of edits

20
Evaluation
Evaluation of this research will take place in
three stages 1. Checklist-based evaluation of
ongoing project - Ensures that goals are
met, identifies risks etc. 2. Initial user
evaluation with MSc students. -
Controllable environment, short, validates design
assumptions identifies problems but
isn't very realistic 3. Full scale evaluation
with as-yet unidentified industrial or academic
partner - Realistic. Intention is to
validate tool in real-world use and
identify goals for future research. The
user-focussed evaluations are questionnaire and
observation based. Additionally the awareness
tools will be instrumented for debugging and data
collection purposes.
21
Open Source ERP for SMEs

• Hyoseob Kim
• Dept. of Computing and Informatics
• Faculty of Technology
• University of Lincoln

1/4
Aston University, 20 April 2005
22
Open Source ERP for SMEs

• Funded by a University of Lincoln Research
  Excellence Grant
• Using open source ERP (Enterprise Resource
  Planning) packages for SMEs (Small to
  Medium-sized Enterprises)
• Fierce competition predicted in the SMEs ERP
  market because of the top-end market saturation
• Building blocks of an ERP system, e.g., OS and
  DBMS can now be open-sourced.
• There is a strong demand and interest but
  efforts need to be more focussed and better
  organised to produce quality software.

Aston University, 20 April 2005
2/4
23
Survey of Open Source ERP Market

• SourceForge.net, the largest open source software
  repository
• Had expected a small number of industrial-strength
  packages
• But found many low-quality packages 129 in
  total, majority of these inappropriate for an
  industrial use
• 75 projects (58) were one-person projects, and
  115 projects (89) have less than or equal to
  five developers.

Aston University, 20 April 2005
3/4
24
Open Source ERP Evolution

• Investigating the developers' various
  characteristics, e.g., why do developers want to
  start their own projects rather than joining
  existing ones?
• Surveying customers' needs Know your
  customers' needs
• Looking into the evolutionary pattern of open
  source ERP packages to see whether it is
  different from other open source software, e.g.,
  Linux and Apache and more conventionally
  developed software (cf., Eight Laws of Software
  Evolution).

Aston University, 20 April 2005
4/4
25
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Towards Supporting Agile Practice Within The
Libre Software Paradigm
Paul J. Adams BSc, MBCS Research Assistant
(CALIBRE)
Research History
2003 2004 Final Project Using Open Source
Components To Support Distributed Software
Development Jun. 2004 Graduate BSc (Hons)
Software Engineering University of Durham,
UK Sept. 2004 Research Assistant EU FP6 -
CALIBRE Project Part-time PhD
candidate University of Lincoln, UK
26
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
To What Extent Can Distributed Agility Improve
Open Source Practice (And Vice Versa)?
This research will take place as part of research
to answer the broader question To What Extent
Can Agile Practice Be Effectively Distributed?
Research Scope

• Agility
• The agile methods, the agile manifesto,
  evaluation of agility, DXP, metricsof
  performance within agile practice
• Open Source
• Process within open source, performance drivers
  in open source

27
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Motivation Pushing The Boundaries of Agility
The Troubles With Agile Practice

• The Agile methods are just a strict as their
  classic counterparts
• The usefulness of Agile practice is limited
• We can infer a requirement of team collocation
• Agile practice is best suited to small
  development teams

The Aim Of This Research
The main aim of this research is to extend the
number of useful applications of agile practice
through the loosening of the collocation and
small development team requirements. In
particular, this research aims to facilitate the
use of agile practices within open source
software development.
28
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Research Approach
Towards A Liberal Paradigm Of Software
Engineering

• Abstraction
• Identifying the process within open source
  projects
• Identifying areas of commonality amongst
  these processesDistillation
• Learning best practice from open source and
  agile
• Producing a well-defined Liberal paradigm

• Software Development
• Creation of new tools to support processes
  within the Liberal paradigm, developed as a
  plug-in for the Eclipse IDE
• Tool set adjusted through experimentation and
  evaluation

29
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Distributed Software Engineering Research Group,
Department of Computing and Informatics,
University of Lincoln
Research Limitations
Tool Set Limitations

• Initial prototype tool support has been focused
  on the eXtreme Programming method
• Pair programming, story cards, collaborative
  white board etc.

Evaluation Limitations

• Evaluation of tool use within open open source
  projects will only give an indication of
  performance within other contexts
• Evaluation of the complete system will be
  limited to a small number of test projects
• There can be no guarantee that the experimenters
  use the tool exclusively