$29.6 KEY CONCEPTS STUDIED IN THIS CHAPTER 947 professors.This assumes of course that the application domain and the breadth of the library's coverage have been chosen judiciously so as to match the size of the available resources in people,equipment and funds. Talking about resources,the project may start with relatively limited means but is a prime candidate to attract the attention of funding agencies.It also offers prospects of industry funding if the application domain is of direct interest to companies. See“APPENDIX: Building good libraries is a technically exciting task,which raises new scientific A HISTORY OF challenges,so that the output of a successful project may include theses and TAXONOMY". 24.15,page864. publications,not just software.The intellectual challenges are oftwo kinds.First the construction of reusable components is one of the most interesting and difficult problems of software engineering,for which the method brings some help but certainly does not answer all questions.Second,any successful application library must rest on a taxonomy of the application domain,requiring a long-term effort at classifying the known concepts in that area.As is well known in the natural sciences (remember the discussion of the history of taxonomy),classification is the first step towards understanding.Developed for a new application area,such an effort,known as domain analysis,raises new and interesting problems. The last comment suggests the possibility of inter-disciplinary cooperation with researchers in various application domains,usually non-software. Cooperation should begin with people working in neighboring fields.Many universities have two groups pursuing teaching and research in software issues,one (often "computing science")having more of an engineering and scientific background,the other(often"information systems")more oriented towards business issues.Whether these groups are administratively separate or part of the same structure-both cases are common-the project may appeal to both,and provides an opportunity for collaboration. Finally,a successful library providing components for an important application area will be widely used and bring much visibility to its originating institution. No doubt in the years to come a number of universities will seize on these ideas,and that the "X University Reusable Financial Components"or "Y Polytechnic Object- Oriented Text Processing Library"will (with better names than these)bring to their institutions the modern equivalent of what UCSD Pascal,Waterloo Fortran and the MIT's X Window system achieved in earlier eras for their respective sponsors. 29.6 KEY CONCEPTS STUDIED IN THIS CHAPTER In object-oriented training,emphasize implementation and design. In initial training for professionals,do not hesitate to repeat a session,with some time in-between for actual practice. Training in a company should include courses for managers as well as developers.§29.6 KEY CONCEPTS STUDIED IN THIS CHAPTER 947 professors. This assumes of course that the application domain and the breadth of the library’s coverage have been chosen judiciously so as to match the size of the available resources in people, equipment and funds. • Talking about resources, the project may start with relatively limited means but is a prime candidate to attract the attention of funding agencies. It also offers prospects of industry funding if the application domain is of direct interest to companies. • Building good libraries is a technically exciting task, which raises new scientific challenges, so that the output of a successful project may include theses and publications, not just software. The intellectual challenges are of two kinds. First the construction of reusable components is one of the most interesting and difficult problems of software engineering, for which the method brings some help but certainly does not answer all questions. Second, any successful application library must rest on a taxonomy of the application domain, requiring a long-term effort at classifying the known concepts in that area. As is well known in the natural sciences (remember the discussion of the history of taxonomy), classification is the first step towards understanding. Developed for a new application area, such an effort, known as domain analysis, raises new and interesting problems. • The last comment suggests the possibility of inter-disciplinary cooperation with researchers in various application domains, usually non-software. • Cooperation should begin with people working in neighboring fields. Many universities have two groups pursuing teaching and research in software issues, one (often “computing science”) having more of an engineering and scientific background, the other (often “information systems”) more oriented towards business issues. Whether these groups are administratively separate or part of the same structure — both cases are common — the project may appeal to both, and provides an opportunity for collaboration. • Finally, a successful library providing components for an important application area will be widely used and bring much visibility to its originating institution. No doubt in the years to come a number of universities will seize on these ideas, and that the “X University Reusable Financial Components” or “Y Polytechnic Object￾Oriented Text Processing Library” will (with better names than these) bring to their institutions the modern equivalent of what UCSD Pascal, Waterloo Fortran and the MIT’s X Window system achieved in earlier eras for their respective sponsors. 29.6 KEY CONCEPTS STUDIED IN THIS CHAPTER • In object-oriented training, emphasize implementation and design. • In initial training for professionals, do not hesitate to repeat a session, with some time in-between for actual practice. • Training in a company should include courses for managers as well as developers. See “APPENDIX: A HISTORY OF TAXONOMY”, 24.15, page 864