Climbing the tree of unreachable fruits, reusing processes
1 like348 views
- The document discusses the history and evolution of software process research from the 1980s to present day, including the development of process modeling languages, standards like CMMI, and agile methods like Scrum. - It notes that while early attempts at fully automated software engineering environments did not succeed, many useful tools were developed for configurations management, testing, project management, and more. - Current challenges discussed include coping with faster development speeds, integrating research ideas into practical tools, educating engineers, and addressing social and technical aspects together.
Climbing the tree of unreachable fruits, reusing processes
- 1. Climbing the tree of unreachable fruits reusing processes P R O F . D R . I T A N A G I M E N E S UNIVERSIDADE ESTADUAL DE MARINGÁ CBSOF T , 2 0 1 4
- 2. Summary — An overview of the software process research ¡ From well-defined process to software quality ¡ From fully-automated processes to toolkits ¡ Is software process really different from workflow management? ¡ The Agile breakthrough ¡ Current Challenges — Reflections on our community ¡ JSERD
- 3. Our first belief, 27 years ago — “Software processes are software too” (Osteweil, ICSE 1987) ¡ Well-defined processes might lead to reductions in software production costs and improvements in software quality. ¡ Software Process is the total set of software engineering activities employed in the production, assessment and certification of an operational system (Gimenes, 1992; Liu & Conradi, 1991). Software Process Model is a formal representation (abstraction)
- 4. From well-defined process to software quality — Process Modelling Languages (PML) ¡ DesignNet (Petri Net – based) ¡ Formal models (considering activity as mathematical functions) ex. HFSP ¡ Rule-based models – activities are defined as rules ex. Marvel ¡ Plan-based – emphasis on goals, ex. GRAPPLE. EPOS PML
- 5. PML went too far and basically missed the point (Fuggeta, 2000)
- 6. From well-defined process to software quality — Process Improvement and Assessment (from mid 80s) ¡ Software process is dynamic ¡ ISO9000s ¡ ISO/IEC 12207, 15504 ¡ IEEE ¡ CMM (1987), CMMI (2001) ¡ SPICE ¡ SPEM ¡ MPS-BR (Estudo de Caso: 10 Anos de MPS.BR’, Softex, 2014) ÷ 2004 – 2007 MPS-BR Implementation ÷ 2008 - 2011 MPS-BR Consolidation
- 7. • A strong community was formed supported by: Academy, Government and Industry. • Evaluation and measurement prevails over prescription and fully automation.
- 8. From fully-automated processes to toolkits — Software Engineering Environment (SEE), 1990s ¡ A set of integrated tools acting over domain-specific database which can support software engineers throughout the software development process. — PSEEs – Process-centered SEEs ¡ A SEE with a top level layer that controls the software process. — PCTE ¡ a framework for tool integration
- 9. The PCTE Framework Process Integration Process Step How well do relevant tools combine to support the performance of a process step? Event How well do relevant tools agree on the events required to support a process? Constraint How well do relevant tools cooperate to enforce a constraint? Tool Provision To what extent are a tool’s services used by other tools in the environment? Use To what extent does a tool use the services provided by other tools in the environment? Control Integration Figure 4-1 Properties of Integration1 Presentation Integration Appearance and Behavior To what extent do tools use similar screen appearance and interactive behavior? Interactive Paradigm To what extent do two tools use similar metaphors and mental models? 4.1 Platform Integration Data Integration Interoperability How much work must be done to manipulate data produced by another? Nonredundancy How much data managed by a tool is duplicated in or can be derived from the data managed by the other? Data Consistency How well do tools cooperate to maintain the semantic constraints of the data they manipulate? Data Exchange How much work must be done to make the nonpersistent data generated by one tool usable by the other? Synchronization How well does a tool communicate changes it makes to the values of nonpersistent, common data? CMU/SEI-93-TR-1 The intent of platform integration is to provide network and operating system transparency to
- 10. From fully-automated processes to toolkits — PSEEs didn’t survived. — However, many good ideas survived into tool kits ¡ Configuration management, ex. SCCS, Git ¡ Software building, ex. Maven, Jenkins ¡ Programing environments, ex. Netbeans ¡ Source code analysis, ex. SonarQube ¡ Testing, ex. XUnity family ¡ Project management, ex. Jira — Integration is still a burden.
- 11. We have learned a lot about how to integrate tools, but the straight jacket failed again!
- 12. The Impact of the Workflow Movement — Questions: ¡ Is software process really different from workflow? ¡ Is the support environment different? ¡ and business process?
- 13. The Impact of the Workflow Movement — Workflow Handbook, 1995 ¡ Workflow – collection of activities organized to realize a business process. ¡ Workflow (automation) x Process (Definition). ¡ Workflow Management Systems (WfMS) – systems that support the definition, creation, management and execution of workflows. ¡ Workflow Management Coalition (WfMC, Similar 1999) ideas and defined a reference model for WfMS (Components and Interfaces). different communities
- 14. The Impact of the Workflow Movement — Workflow community created a prosperous research field which led to the Business Process Management community. — The Workflow Reference Model 10 years on (WfMC, 2005): “It has been a subject of some debate whether there is any practical difference between workflow management and business process management. Certainly many of the concepts are the same and, where there are differences, these tend to be in points of detail, or through different emphasis” ¡ BPM evolves from Workflow + EAI + Web
- 15. When Business = Software, software process is equal to Business Process. High level languages and tools, such as BPMN can be applied to specify and simulate software processes.
- 16. The Agile Breakthrough, 2001 — Questions: ¡ Is the process an end in itself? ¡ It is too rigorous, is it effective? ¡ Does it deal with human behavior? — Values: Individuals and interactions over processes and tools Working software over comprehensive documentation Customer collaboration over contract negotiation Responding to change over following a plan
- 17. The Agile Breakthrough, 2001 Principles — the software product is the essential focus of the development process; — human behavior and quality of interactions are essential enablers and success factors; — incremental and spiral approaches, based on frequent releases and strict collaboration with the costumer are quintessential traits of modern software development initiatives; — in general, it is vital to promptly and quickly react to requirement changes, at any stage of the development process.
- 18. The Agile Breakthrough,2001 — XP ¡ Code ¡ Test-driven development ¡ Pair programming — Project Management ¡ SCRUM ¡ Lean ¡ KanBan
- 19. “Nobody can tell you the state of agile; Everything we know is subjective and based on circumstantial evidence” (Ken Schwabe, 2014)
- 20. The impact of the Open Source Movement — The Mozilla Foundation (Mitchell Baker) — Brought challenges in ¡ Distribution ¡ Flexibility ¡ Ability to deal with change requirements
- 21. We have expanded software development a lot, but this does not apply for all types of software.
- 22. Reflection ….
- 23. Current Challenges - not only research — Coping with the technology push: Internet, services, mobile, … ¡ We had a belief that we should elicit requirements and then make design decisions but the available technology is driving the process. — Coping with the increasing speed push ¡ There seems not to be enough time for careful planning and thinking; ¡ Time-to-market has been more important than reliability/ quality.
- 24. Current Challenges - not only research — Turn ideas into tools, frameworks, components … ¡ Our research ideas only get to practice when there is proof that to be feasible, operational. ¡ software engineering has to complete its cycle of ideas, develop tools, develop experiments. ¡ research evaluation mechanisms impair this cycle, they give too much value to novel ideas. ¡ Model-driven engineering can help.
- 25. Current Challenges - not only research — Educate software engineers with both technical and soft skill ¡ Technical background: computer science, mathematics, software methods; ¡ Deal with the market distance; ¡ Stimulate software skills: creativity, work in group, problem solving, leadership.
- 26. Current Challenges - not only research — Bring technology and social aspects in context — Team distribution (Global Software Development) — Empirical studies are not an end in itself ¡ Limitations: participants, scenarios, data volume. — Flexible control ¡ Application Lifecycle Management (ALM) (Fuggeta 2014), ex. MyLin, Tasktop ¡ Smart convergence (Fuggeta, 2014)
- 27. • Research on software process and support environment have produced many fruits; not all of them worked. • There are many challenges to face, the main message • Evaluate your research, try to be proud of what you do, not only of producing another paper.