process and open source software walt scacchi institute for software research uc irvine...
TRANSCRIPT
Process and Open Source Software
Walt ScacchiInstitute for Software Research
UC Irvine
http://www.ics.uci.edu/~wscacchi/Presentations/Process/Informatix-Process-Lecture.ppt
Overview
Core concerns Software process and process modeling
Backstory/case study History of software process Current challenges in software process Open source software processes Best practices
Core concerns Process(es)
What to do and how to do it Tasks, inputs, outputs, performers, tools
Performers enact tasks using tools to transform inputs into outputs
Tasks are partially ordered Tasks, inputs, performers, tools may have pre-conditions Tasks, outputs, performers may have post-conditions
Organize and coordinate who does what, (with what resources), when, where, how, and why
Formal modeling of processes enables systematic querying and reasoning about complex processes
Core concerns Process modeling
Understanding, analyzing, and (re)designing software development processes
Models may be either Prescriptive—what should be done and how to do it Proscriptive—what might be done and how Descriptive—what was done and how (historic record)
Models may be either Narrative Semi-structured (form-based; hypermedia-based) Graphic (attributed flow graphs) Formal (language-based process programs or specifications)
Families of models inter-related through meta-model Meta-model provides model ontology and process epistemology
Backstory/case study Major TelCo wants to develop broadband multi-media
telecommunications system Anticipates $1B software development, up to 1500
software developers working 2-3 years Seeks industrial/IT partners to provide supporting
infrastructure to reduce risk IT partner wants to showcase new “process support
technology” products as sales lead IT partner brings in academic research team to analyze
and advise TelCo on “process issues”
Backstory
Academic team, IT partner, and major TelCo jointly elicit, capture, codify (model) and inter-relate TO-BE system development process.
Academic team employs IT partner’s products to present results of their “process analysis”
IT partner considers its process-centered product major success to help large sales engagements
Academic team view of their effort -- a major success for publication, but...
P.K. Garg, P. Mi, T. Pham, W. Scacchi, and G. Thunquest, The SMART Approach to Software Process Engineering, Proc. 16th. Intern. Conf. Software Engineering, IEEE Computer Society, Sorrento, Italy, 341-350, May 1994.
A complex organizational process (and model): a decomposition-precedence relationship view
(19 levels of decomposition, 400+ tasks)
W. Scacchi, Experience with Software Process Simulation and Modeling, J. Systems and Software, 46(2/3):183-192,1999.
Backstory
Academic team suggests overall process will not succeed -- too complex, too much delegation, problematic hand-offs (“throwing it over the wall”)
TelCo and IT partner dismiss academic team Less than one year later, IT vendor abandons process
technology product Two years later, business press reports TelCo
experiences major project failure and losses greater than $200M, and no system.
Software process model history
Process flow charts Software life cycle (“waterfall” diagram) Incremental development Spiral model Process programs (formal specification) Capability maturity models and industrial/military
standards Not process models--only focus on what to do, but not
how to do it.
Recent software process challenges
Distributed, decentralized, and/or global software development
Process improvement Process design optimization or redesign Continuous process improvement (learning)
Articulating software evolution processes Open source software development processes Process discovery
(semi-)automated extraction of process events, conditions, timestamps, and other meta-data from software development artifacts
J. Noll and W. Scacchi, Supporting Software Development in Virtual Enterprises, Journal of Digital Information 1(4), February 1999.
What is free/open source software development?
Free (as in “freedom”) vs. open source Freedom to access, browse/view, study, modify and
redistribute the source code Free is always open, but open is not always free
F/OSSD is not “software engineering” Different: F/OSSD can be faster, better, and cheaper than
SE in some circumstances F/OSSD involves more software development tools, Web
resources, and personal computing resources
OSS Development Models
Free Software (GPL) Open Source (BSD/MIT, Mozilla, Apache) Corporate Source (Hewlett-Packard) Consortium/Alliance (OSDL, SugarCRM) Corporate-Sponsored (IBM-Eclipse, Sun-Netbeans,
Sun-OpenOffice, HP-Gelato) Community Source (Sakai, Westwood)
OSSD Project Characteristics
OSS Developers are always users of what they build, while OSS users (>1%) are also OSS developers
Requires “critical mass” of contributors and OSS components connected through socio-technical interaction networks
OSSD projects emerge/evolve via bricolage Unanticipated architectural (de)compositions Multi-project component integrations
OSSD teams use 10-50 OSSD tools to support their development work
OSSD Project Characteristics
Operational code early and often--actively improved and continuously adapted
Post-facto software system requirements and design OSSD is not Software Engineering OSSD has its own “-ilities” which differ from those for
SE Caution: the vast majority of OSSD projects fail to
grow or to produce a beta release.
OSS Processes for Requirements or Design
OSS Requirements/Designs not explicit not formal
OSS Requirements/Designs are embedded within “informalisms”
Example OSS informalisms to follow (as screenshot displays)
OSS Requirements/Design processes are different from their SE counterparts.
SE vs. OSS processes for Requirements• Elicitation• Analysis • Specification and
modeling• Validation
• Communicating and managing
• Post-hoc assertion• Reading, sense-
making, accountability• Continually emerging
webs of discourse• Condensing and
hardening discourse• Global access to
discourse
Evolutionary redevelopment, reinvention, and redistribution in OSS One evolutionary dynamic of OSSD is reinvention
Reinvention enables continuous improvement OSS evolve through minor mutations
Expressed, recombined, redistributed via incremental releases OSS systems co-evolve with their development
community Success of one depends on the success of the other
Closed legacy systems may be revitalized via opening and redistribution of their source When enthusiastic user-developers want their cultural experience
with such systems to be maintained.
Configuration management and work coordination in OSSD
Use CM to coordinate and control who gets to update what part of the system/online artifacts. Many OSSD projects use CVS (single centralized code
repository with update locks) and frequent releases (daily releases on active projects)
Linux Kernel: Git (multiple parallel builds and release repositories)
Collab.Net and Tigris.org: Subversion (CVS++) Apache: Single major release, with frequent “patch”
releases (e.g., “A patchy server”) GNU arch seeks to develop Free CM unification
Project management and career development in OSSD
OSSD projects self-organize as a meritocractic role-hierarchy and virtual project management Meritocracies embrace incremental innovations over radical
innovations VPM requires people to act in leadership roles based on skill,
availability, and belief in project community
OSS developers value freedom of choice and expression. want to learn about new stuff (tools, techniques, skills, etc.), have fun building software, exercise their technical skill (e.g., developing reusable code), try out new kinds of systems to develop, and/or interconnect multiple F/OSSD projects
Socio-technical and cultural evolution processes in OSSD
New kinds of OSSD processes under study Joining and contributing to a project in progress Role-task migration: from project periphery to center Alliance formation and community development
Independent and autonomous project communities can interlink via social networks that manipulate objects of interaction Enables possible exponential growth of interacting and
interdependent community as socio-technical interaction network
Discovering the what and how of OSSD processes
Practitioner reflections and anecdotes Surveys Ethnographic field work (virtual ethnography) Mining OSSD project repositories Multi-modal modeling, analysis, and validation of
OSSD processes
Sun Microsystems
The BoardRelease Manager
Maintainer Developers/ Contributors
Website
Mailing Lists
Users
Contribute to community
, meet time constraints for the release
Maintain a project/ module, manage a group of developers
Ensure that the netbeans
community is being run in a fair and open manner
Start new release phase, propose schedule/plan
CVS
Funds, support, Promote Java/Open source
IssueZilla
QA Team
Produce Q- builds and ensure quality of the software
release proposal, release updates,
branch for current release, release post mortem,
review release candidates (2) & decide final release
download development builds and test
, release Q-builds
download new release
report bugs
select feature to develop
, bug to fix, download netbeans,
commit code
decide features for the project and merge patches/bug fixes
, create module web page
make decisions for the community, on high level
grant CVS commit privilege to developers
Link to all Use Cases Links to all Agents
CVS Manager
Configure and maintain CVS
grant access
Site Administrator
Manage website
deploy builds
SourceCast
Tools
Link to Tools
Download and use free software
Community Manager
Share knowledge and ensure all community issues are addressedrespond to tech issu
es, unanswered questions
Rich Picture
Formal model of an OSSD process coded in PML (excerpt) ... sequence Test { action Execute automatic test scripts { requires { Test scripts, release binaries } provides { Test results } tool { Automated test suite (xtest, others) } agent { Sun ONE Studio QA team } script { /* Executed off-site */ } } action Execute manual test scripts { requires { Release binaries } provides { Test results } tool { NetBeans IDE } agent { users, developers, Sun ONE Studio QA team, Sun ONE Studio developers } script { /* Executed off-site */ } } iteration Update Issuezilla { action Report issues to Issuezilla { requires { Test results } provides { Issuezilla entry } tool { Web browser } agent { users, developers, Sun ONE Studio QA team, Sun ONE Studio developers } script { <br><a href="http://www.netbeans.org/issues/">Navigate to Issuezilla </a> <br><a href="http://www.netbeans.org/issues/query.cgi">Query Issuezilla </a> <br><a href="http://www.netbeans.org/issues/enter_bug.cgi">Enter issue </a> } } …
PML validation analysis
Summary of analysis for netbeans_req_release.pmlModel size (source lines): 307Actions: 36Resources: 72Actions neither requiring no r providing resources: 1Resources required but not provid ed (potential inputs): 0Resources provided but not required (potential outputs): 0Miracles: 2Black holes: 6Transfo rmations: 30
Best practices
Processes with explicit process models are easier to manage, analyze, improve, distribute, and reuse
New/ unfamiliar software tools and techniques are best candidates for software process support
Process meta-modeling enables process life cycle engineering and formal reasoning about processes
OSSD is a community building process not just a technical development process F/OSS peer review creates a community of peers
OSSD processes often iterate daily versus infrequent singular (milestone) Software Life Cycle Engineering events OSSD: frequent, rapid cycle time (easier to improve) vs. SLC: infrequent, slow cycle time (harder to improve)
Best practices
Easiest to improve a process that is formally modeled
Process management and improvement have been one of the most enduring practices in Software Engineering for improving productivity and quality, and to reducing cost and risks.
Research opportunities
FOSSD is poised to alter the calculus of empirical SE Software process discovery, modeling, and
simulation Repository mining can support software
visualization, refactoring/redesign studies Comparison of SE versus FOSSD approaches to
software inspection and peer review
Research opportunities
Based on results from individual motivation, participation, role migration, and turnover in FOSSD projects, SE world would benefit from empirical studies that examine similar patterns in conventional software development projects Is FOSSD more fun, interesting, and rewarding than
SE?
Research opportunities
Conventional software cost estimation techniques (e.g., “total cost of operation”) slight/ignore social capital and socio-technical resources Miscalculation of total resources and capabilities
that affect predicted/actual costs of software development or FOSSD
Research opportunities
Results from study of cooperation, coordination and control in FOSSD Virtual project management and role migration
processes can provide a lightweight approach to SE project management
Unclear whether proprietary software projects willing to embrace VPM
Research opportunities
Alliance formation and social networking results suggest SE projects operate at a disadvantage compared to FOSSD projects SE projects tend to produce systems whose
growth/evolution is limited FOSSD projects can produce systems capable of
sustained exponential growth/evolution of both software and developer-user community
Research opportunities
How best to encourage the emergence of a social movement that combines best practices of FOSSD and SE Consider participation or study of open source
software engineering (OSSE) projects at Tigris.org OSSE seeks to combine SE and FOSSD tools,
techniques, and concepts
Detailed study report available
W. Scacchi, Free/Open Source Software Development: Recent Research Results and Methods, in M. Zelkowitz (ed.), Advances in Computers, Vol. 69, 243-295, 2007.
• http://www.ics.uci.edu/~wscacchi/Papers/New/Draft_Chapter_Scacchi.pdf
Acknowledgements
Project collaborators: Mark Ackerman, UMichigan, Ann Arbor Les Gasser, UIllinois, Urbana-Champaign John Noll, Santa Clara University Margaret Ellliot, Chris Jensen, UCI-ISR and others ar ISR
Funding support: National Science Foundation: #0083075, #0205679, #0205724,
#0350754 and #0534771. Digital Industry Promotion (DIP) Agency, Global R&D Collaboration
Center, Daegu, South Korea No endorsement implied.