From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.5-pre1 (2020-06-20) on ip-172-31-74-118.ec2.internal X-Spam-Level: X-Spam-Status: No, score=-0.0 required=3.0 tests=BAYES_40 autolearn=ham autolearn_force=no version=3.4.5-pre1 Date: 14 Apr 93 13:58:01 GMT From: enterpoop.mit.edu!spool.mu.edu!howland.reston.ans.net!noc.near.net!inmet! dsd!ryer@ucbvax.Berkeley.EDU (Mike Ryer) Subject: Re: Ichbiah's Letter Message-ID: <1993Apr14.135801.23812@inmet.camb.inmet.com> List-Id: The Ada 9X Mapping/Revision team is in a Distinguished Reviewers meeting this week, and probably will not see the request for a response to Ichbiah's letter until next week. I think they will respond. Meanwhile, since I sometimes wear the Intermetrics Ada Marketing hat, I thought I'd contribute some genuine "words from marketing" to this discussion. You may find a partial response to some of Dr. Ichbiah's points herein, but bear in mind that this is NOT the official mapping team response -- it's just my opinion. Here are some things I will and do say in a marketing capacity: 1. The major new features in Ada 9X are OOP, tighter realtime, and hierarchical libraries. 2. 9X has real OOP, with inheritance and polymorphism. Ada 9X classes are not identical to the C++ or SmallTalk versions, but provide the same capability. The major distinction is that in Ada 9X, modularity is independent from hierarchy. In Ada 9X, OOP is well-integrated with the other language features, and all of the existing software engineering facilities work smoothly with OOP. Ada 9X is an excellent OOP language, even for applications where maintainability and safety are not overriding issues. 3. Realtime programmers needed facilities for synchronizing that were more predictable and had lower overhead. These facilities are provided by Ada 9X Protected Types. Protected types can be implemented by very efficient techniques that avoid process swaps and queueing. 4. Hierarchical libraries reduce the compilation time requirements for large systems, and provide another effective mechanism for hiding details of implementations from the users of those implementations. 5. The addition of OOP features, and the use of hierarchical libraries to reduce compilation times, will make Ada 9X highly competitive with C++ and, in my opinion, much better than C++. 6. Dr. Ichbiah's diatribe against Ada 9x is unfortunate. His view that OOP should be the only major extension has been outweighed by the demands of users for improvements. Every proposed change in Ada 9X was requested by the user community, and the general applicability and importance of the request was carefully weighed through a open requirements analysis process. The Mapping/Revision team is proud of having meet all of the requirements with so few new constructs and with so few incompatible changes. Intermetrics made several proposals to further simplify the Ada language, by removing complexities and exotic special cases from the base (Ada '83) language. However, the 9X team decided that upwards compatibility was more important than simplifying the base language, and these simplifying proposals were dropped. Ada 9X adds only moderate complexity, and is extremely upward compatible. 7. We will be marketing Ada 9X into markets outside the DOD. For example, Ada 9X will have a booth at OOPSLA. However, in the commercial marketplace it is very inappropriate to start selling products before they are available. We will be at Object World and similar conferences as soon as we can promise an availability date for Ada 9X compiler systems. (You may note that Intermetrics is already demonstrating a class browser for Ada 9X as a technology -- we will be demonstrating it as a product soon). 8. Perhaps the generals who are now saying "Ada hasn't worked, lets do somethin g else" are the ones who never gave Ada a try. They dodged the mandate for years, and now are unable to stay in the closet with their other languages, so they are trying to fight now. If I owned the Defense Department ;-), I'd order them to shut up until they have done all of their programs in Ada for three years; if they still have a problem then, I'd listen. In a way, I consider the currrent outcry a positive sign; it means that the mandate is finally beginning to affect the whole DOD, and not just the highly visible major weapon systems. 9. I acknowledge that it is very hard to learn Ada 9X from the current documentation. The forthcoming reference manual will be usable by a wider range of engineers than the current ILS or the earlier mapping documents. John Barnes is writing a Rationale which is still more readable, and there will be training tapes made as well. Eventually the textbooks, on-line tutorials, training programs and other introductory material will be available. We'll need all of this stuff as soon as compilers hit the street but until then, concentrating on the full formal semantics in a reference format is the best way to pin down the details of the revision. -- Mike "removing marketing hat, donning flame-proof suit" Ryer Intermetrics, Inc.