Master Thesis Project Plan. Reusable Mathematical Models

Transcription

1 Master Thesis Project Plan Reusable Mathematical Models Tobias K. Widmer Supervisors: Prof. Dr. B. Meyer B. Schoeller Chair of Software Engineering Department of Computer Science, ETH Zurich January 13, 2004

2 Chapter 1 Project Description 1.1 Overview Proving programs in a fully mathematical and rigorous way is hard and time consuming. Although there exist a number of formal approaches to software engineering like the Z specification language or the B method, these environments strongly focus on refinement based programming and do not directly support object-oriented programming. On the other side, the JML specification language is an object-oriented extension to the JAVA programming language, which allows annotating existing JAVA programs with specification attributes and contracts. But neither of these approaches really take advantage of the most beneficial concept in software engineering: Reuse of already available software elements. It should not only be possible to reuse units of implemented software, but also rely on units of already proven software. Unfortunately, the traditional methods like B and JML only gradually support the reuse of already proven software units. In the Eiffel method, contracts have become first-class citizens of the programming language. Designed as taking an integral part in the software specification process and providing expressive semi-formal definitions of software interfaces, it turns out that Eiffel contracts are not powerful enough to express the complicated relationships in object graphs of modern object-oriented environments. Faced with these shortcomings, the idea to combine the reusability and extendibility of object-oriented methods with the rigorous mathematical models used in formal method like B seams reasonable. The idea of applying mathematical models to prove contract equipped classes has been developed by [Mey03]. The open question remains how these mathematical concepts can be modeled as software components to help making Eiffel contracts even more expressive and powerful. The goal of this thesis is therefore to create a collection of extendible and reusable software components that allow Eiffel programmers to write more rigorous and expressive contracts using abstract mathematical models. 1.2 Scope of the work Translating mathematical concepts into reusable software models requires several subproblems to be solved. In the first part of this thesis, the basic blocks of mathematical reasoning over programs have to be identified and categorized. We therefore give an overview about the necessary mathematical concepts and justify their necessity with respect to the aim of providing a minimal, yet expressive enough set of software facilities to describe Eiffel contracts. The approaches taken in B and Z are certainly influencing the result. The ideas behind JML are considered as well, since JML promises to be much closer to object-oriented methods than the other approaches. After the necessary mathematical concepts have been identified, a representative set of data structures 1

3 has to be chosen to deliver the proof of concept, meaning that these data structures can be specified using the acquired models. For us, the EiffelBase library serves this purpose. Only the most fundamental data structures are specified. As time permits, more advanced data structures are treated as well. This specification phase is followed by an implementation phase, during which we create the necessary software facilities to support the specification of the above data structures. This collection of software classes will be called Mathematical Model Library (MML) and is written in a defined subset of the Eiffel language, which we name Intermediate Functional Language (IFL). This language provides the client programmer with the necessary constructs to define expressive contracts and is restricted to use features from the MML class library only. 1.3 Intended results Mathematical Model Library The library of mathematical models forms the most important deliverable. It will comprise the following items: 1. A library of thoroughly documented classes modeling the mathematical concepts identified in the first part of this thesis 2. A user s manual describing the scope and usage of the library classes Thesis Report The report associated with this thesis covers the following topics: 1. An overview about the existing formal approaches to software engineering. We introduce the concepts and ideas behind B and JML and how they could be used in the context of Eiffel contracts 2. A compilation of basic mathematical concepts used for the design of the MML class library 3. Mathematical models of the most fundamental data structures contained in the EiffelBase library 4. The specification of the functional language IFL needed by the MML classes 5. The design ideas and architectural decisions behind the MML class library In the end, this master thesis should result in reusable software components that are ready to be used in contracts of programs written in the Eiffel language. 2

4 Chapter 2 Background Material 2.1 Reading List In this chapter we introduce the tentative list of bibliographic sources used in this thesis: Formal Methods: The B-Book [Abr96], The B-Method: An Introduction [Sch01] The Z notation [Spi89] JML Reference Manual Draft [LPC + ], JML: A notation for detailed design [LBR], Design by Contract with JML [LC] Program Verification: Towards practical proofs of class correctness [Mey03], A logic for the Verification of Object- Oriented Programs [PHM97], Formal Verification of a Doubly Linked List Implementation: A Case Study Using the JIVE System [LMMPH00] Mathematical Concepts: Introduction to the Theory of Programming Languages [Mey90], The Z notation [Spi89] Eiffel Language: Object-Oriented Software Construction [Mey97], Eiffel: The language [Mey92] Software Engineering Design Patterns: Elements of Reusable Object-Oriented Software [GHJV95], Introduction to Algorithms [CLRS01] This list is certainly not exhaustive and is likely to be extended during the course of this master thesis. 3

5 Chapter 3 Project Management 3.1 Objectives and priorities The primary objective of this thesis is to deliver in a timely fashion a powerful collection of Eiffel classes that can be used to write expressive and precise contracts. 3.2 Criteria for Success Mathematical Model Library The following criteria have to be met by the implementation of this class library: Extendible and reusable library classes that directly model the needed mathematical concepts A carefully designed class hierarchy which is easily understandable Well documented code with precise and rigorous contracts Easy to use class interfaces that hide the internal mathematical complexity Thesis Report The thesis report should complement the work of the class library: It should give a comprehensive overview about the mathematical concepts necessary to define the intermediate functional language IFL The report should present the specification of the EiffelBase data structures in a clear and concise way addressing Eiffel programmers primarily It has to provide a rigorous specification of the functional language IFL It should contain a broad presentation of the design ideas behind the MML class library 4

6 3.3 Method of work The thesis report and the library classes are created in parallel, hopefully influencing each other positively. The development of the source code is guided by the Seamless Development Process described in [Mey97]. The use of a version control system facilitates the source code management both at home and at work. Frequent validation steps ensure constant progress and minimal deviation from the project plan. 3.4 Quality Management Validation Steps The following validation steps are planned to perform regularly during the course of this thesis: Frequent discussion meetings with the assigned supervisors Weekly reviews and wrap-ups to ensure consistent progress Monthly position-fixings to avoid loosing the plot Test Cases For the software-related part of this thesis, the MML class library, we additionally provide example programs to display the successful application of the mathematical models to Eiffel contracts. This demonstrative examples are accompanied by test cases to ensure a high level of software quality Documentation The class library is fully documented using Design by Contract T M and the Eiffel method style guide lines. The user s manual describes the usage of the library classes from the client programmer s point of view. Information about the design ideas and the architecture of the MML class library can be found in the thesis report. 5

7 Chapter 4 Project Schedule 4.1 Deadlines Project Start: Mo, Project End: Fr, Gantt Chart ID Task Name Duration Start Finish 1 Identify Mathematical 20 days Mon Fri Concepts 2 Reading 5 days Mon Fri Collecting Concepts 4 days Mon Thu Writing Report Chapter 11 days Fri Fri Specify Data Structures 25 days Mon Fri Examine EiffelBase 2 days Mon Tue Create Collection 3 days Wed Fri Write Specification 15 days Mon Fri Write Report Chapter 5 days Mon Fri Define IFL 25 days Mon Fri Review Identified Concepts 5 days Mon Fri Create IFL Draft 10 days Mon Fri Verify Draft on Specifications 4 days Mon Thu Write Report Chapter 6 days Fri Fri Implement MML 35 days Mon Fri Design Class Hierarchy 3 days Mon Wed Specify Interfaces 7 days Thu Fri Implement Classes 15 days Mon Fri Test Classes 5 days Mon Fri Write Examples 5 days Mon Fri Finalization 25 days Mon Fri Write Design and Architecture 6 days Mon Mon Write User's Manual 4 days Tue Fri General Clean-Up 3 days Mon Wed Proof Reading 6 days Thu Thu Reserved 6 days Fri Fri Dec '0 05 Jan '0 19 Jan '0 02 Feb '0 16 Feb '0 01 Mar '0 15 Mar '0 29 Mar '0 12 Apr '0 26 Apr '0 10 May '0 24 May '0 07 Jun '0 21 Jun '0 05 Jul '04 6

8 Bibliography [Abr96] J. R. Abrial. The B-Book. Cambridge University Press, [CLRS01] T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein. Introduction to Algorithms, Second Edition. MIT Press, [GHJV95] E. Gamma, R. Helm, R. Johnson, and J. Vlissides. Design Patterns: Elements of Reusable Object-Oriented Software. Addison-Wesley, [LBR] G. T. Leavens, A. L. Baker, and C. Ruby. JML: a notation for detailed design. ftp://ftp.cs.iastate.edu/pub/leavens/jml/jmlkluwer.pdf. [LC] G. T. Leavens and Y. Cheon. Design by Contract with JML. ftp://ftp.cs.iastate.edu/pub/leavens/jml/jmldbc.pdf. [LMMPH00] M. Labeth, J. Meyer, P. Müller, and A. Poetzsch-Heffter. Formal verification of a doubly linked list implementation: A case study using the JIVE system. Technical Report 270, Fernuniversität Hagen, [LPC + ] G. T. Leavens, E. Poll, C. Clifton, Y. Cheon, C. Ruby, D. Cok, and J. Kiniry. JML Reference Manual Draft. ftp://ftp.cs.iastate.edu/pub/leavens/jml/jmlrefman.pdf. [Mey90] B. Meyer. Introduction to the Theory of Programming Languages. Prentice Hall, [Mey92] B. Meyer. Eiffel: The Language. Prentice Hall, [Mey97] B. Meyer. Object-Oriented Software Construction. Prentice Hall, 2nd edition, [Mey03] [PHM97] B. Meyer. Towards practical proofs of class correctness. In ZB 2003 (D. B. et al., ed.), LNCS 2651, pp Springer-Verlag Berlin, A. Poetzsch-Heffter and P. Müller. A logic for the verification of object-oriented programs. In R. Berghammer and F. Simon, editors, Programming Languages and Fundamentals of Programming, volume 9717 of Technical Report. Christian-Albrechts-Universität Kiel, [Sch01] S. Schneider. The B-Method: An Introduction. Palgrave, [Spi89] J. M. Spivey. The Z notation. Prentice Hall International,