Search

In this paper, we have dealt only with the most characteristic combinations of paradigms and life cycles in Ada. Other combinations are possible: for example, top-down design and bottom-up implementation, or large-small division into large packages first and then bottom-up implementation via subpackages. Since every paradigm/life-cycle combination displays a unique combination of attributes and respective advantages and disadvantages, the project manager in each case should carefully review the circumstances of the project and select the most appropriate combination. Some examples of appropriate paradigm selection are given below: An exploratory program is to be implemented by an individual researcher. Because of the exploratory nature of the program, no requirements specifications exist. Hence, the most appropriate combination is the incremental bottom-up combination, which offers the highest degree of forgiveness (i.e., easy correctability of wrong decisions) and the lowest volume of necessary documentation. The inherent disadvantages of the method chosen are well compensated for in this case: There is only one researcher and therefore no need for parallel effort, and the high professional caliber of the researcher compensates for the greater difficulty of the paradigm. An existing program is to be rewritten in Ada by a team of programmers. The appropriate paradigm for this situation is the large-small traditional combination. It allows several programmers to proceed with implementation in parallel, and its disadvantages are reduced by the fact that the old program already exists and is available to the programmers. Since the experience and design from the old program can be used, the method's lack of forgiveness and high volume of necessary documentation are less relevant. A program in a well-known application area is to be produced by a small team of programmers. In this case, the most appropriate combination is the top-down semi-incremental paradigm as it offers the best certainty of compliance with specifications. Its less forgiving nature is compensated for by the fact that the application area is well known to the programmers; the low level of parallelism of the method is less relevant as a small team is involved. [ABSTRACT FROM AUTHOR]