The ADA/ED system: a large-scale experiment in software prototyping using SETL

  • Authors:
  • Philippe Kruchten;Edmond Schonberg

  • Affiliations:
  • New York Univ., New York;New York Univ., New York

  • Venue:
  • Technology and Science of Informatics
  • Year:
  • 1984

Quantified Score

Hi-index 0.00

Visualization

Abstract

In my view, the most interesting aspect of thisexperiment with prototypes is its implications for software engineering:everyone talks about prototypes today, but is anyone really producingany? Or, amongst those who do produce prototypes, how many can resistthe temptation, once the prototype is complete, to treat it as thefinished product? The experiences described here present a clear cutcase: it would have been impossible to write the compiler immediately inits final form—the cost of perpetual modifications, in response torevisions to ADA, of a compiler written in a `traditional' languagewould have been intolerably high—designing a prototype seems tohave been the only way out.Finally, once a decision in favor of prototypes has been taken,either as a way of giving concrete expression to a set of specificationsor a way of `protecting' a particular design, a choice of language inwhich to write the prototype has to be made. The advantages of SETL areclearly demonstrated in this article, with particular stress laid on thefacilities it provides for using several semantic levels, and thereforerefining a prototype in successive stages until the final productemerges, or a program very close to the final product. Translation intoa language providing for a higher performance level, or at any rate intoa more common language, C in this case, is then a simple task. My guessis that a lot of TSI readers are going to want more information on SETLto enable them to use it in similar circumstances in the future.To sum up, this article is particularly interesting for threereasons: it provides a closer understanding of ADA, demonstrates thevalue of prototypes in software design, and makes a plea in favor ofSETL which does indeed appear wholly satisfactory as a language in whichto write such prototypes.—From the Commentary by E. Girard—Commentary recommended by E. A. Feustel, Sherborn, MA