From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: *** X-Spam-Status: No, score=3.2 required=5.0 tests=BAYES_05,MISSING_DATE, MISSING_MID,MISSING_SUBJECT autolearn=no autolearn_force=no version=3.4.4 From: utzoo!decvax!harpo!floyd!cmcl2!lipkis Newsgroups: net.lang.ada Title: The NYU Ada Project Article-I.D.: cmcl2.30525 Posted: Thu May 12 17:56:53 1983 Received: Thu May 19 02:23:02 1983 List-Id: Message-ID: <20200615230049.GPFs9s23hbJuXG9EM_R0FWXl5noNdM87UxDfojB0oVo@z> A few notes about NYU Ada/Ed: Ada/Ed is indeed very slow. This is by design, and has nothing to do with the use of SETL per se. In fact SETL running code of a PASCAL semantic level runs only 5-10 times slower than PASCAL, even faster in some favourable cases. The point about Ada/Ed is that the semantic level of the interpretor is very high. The actual code of the interpretor is more akin to a formal semantic definition than a conventional interpretor, and indeed Ada/Ed is best regarded as an executable formal definition of Ada, rather than a programming tool. For example, while loops are executed by duplicating copies of the code, and code generation for expressions is done essentially dynamically by walking the expression tree at run time. The entire object of the development is to produce a system with highly readable code (copies of the listing of the interpretor are available from NYU Ada Project, Courant Inst., 251 Mercer St., N.Y.C., NY 10012 at a cost of $25 if you want to see for yourself). Despite this approach, it is possible to use Ada/Ed for familiarization with Ada, and the fact that it has been officially validated is a useful guarantee that someone using Ada/Ed is familiarizing themselves with a reasonably accurate implementation of the language. We have so far strenuously avoided the slightest hint of efficiency, and the whole project has been developed under the conceptual guideline that "slow is beautiful". However, Ada/Ed is in wide use as a training and familiarization tool at several hundred sites, and it appears that there will be an appreciable window in which it will be the only full Ada system available for many machines. Consequently, it is at this stage useful to work on improving its efficiency, and our current project funding includes resources for this purpose. We expect that a version of Ada/Ed will be released later this year with very great execution speed improvements. Robert Dewar (reply to cmcl2!acf2!dewar) NYU Ada Project Courant Inst. of Math. Sci. 251 Mercer St. New York, N.Y. 10012