|
In This Edition:
DoD to Replace Ada Mandate with Software-Engineering Process
AJPO Chief on Special Assignment from West Point
ADEPT – Building Distributed Systems in Ada 95
Updates for Educators: Conferences, Free Courseware, New Books
Ada Diversifies Westinghouse's Czech Nuclear Shutdown System
|
 |
On Nov. 1, 1996, the National Research Council (NRC) at the National Academy of Sciences (NAS) released the draft of its recommendations on the use of the Ada programming language in the Department of Defense (DOD). In mid-January, the final version was published and submitted to the DOD. The NRC report, titled “Ada and Beyond: Software Policies for the Department of Defense”, made three key recommendations: 1) except for what it termed “warfighting” software, DOD should drop its mandate to use Ada; |
|
2) DOD should invest roughly $15 million per year in Ada support in the warfighting domain or drop the Ada requirement completely; and 3) DOD should establish a software-engineering review process for DOD software projects, within which each project would make its language decisions.
(For more on the report itself, see the previous issue of this newsletter. For information on obtaining a copy of the report, one place to start is the Ada Information Clearinghouse’s World Wide Web site (http://archive.adaic.com/docs/present/nrc/). You can also see the NRC’s Web site (http://www2.nas.edu/cstbweb/21b6.html), or contact them at 800/624-6242 (Monday-Friday, 8:30am-5:00pm ET).
Engineering process rather than mandate, even for “warfighting” On Feb. 27, 1997, Hon. Emmett Paige, Jr., Assistant Secretary of Defense for Command, Control, Communications, and Intelligence (ASD C3I), commented on the report in electronic mail to DOD senior executives working in the area of software. He said that he was “prepared to accept and implement all of [the NRC] recommendations with one exception”. Whereas the NRC report had suggested retaining the Ada mandate for “warfighting” systems, Mr. Paige said he believed the DOD “should no longer require Ada for any of its systems....” At the NRC public hearing and in other contexts, questions were raised about the NRC report’s definition of “warfighting” and the DOD’s ability to give this term a clear reference. The NRC study team chose a relatively narrow definition for the term – specifically noting that its use of the term was narrower than widely used terms such as “mission critical”, or “C3I”. In the area of weapons control, for instance, the NRC’s term “warfighting” did not include “support subsystems performing mainstream data management, networking, and graphical user interface functions”. Similar exclusions were noted for other domains. The software-engineering process For the DOD, the primary need for Ada has always been Ada’s inherent support for the principles of sound software engineering. To accomplish these ends without a language mandate, the report recommended that the DOD implement what it termed a software engineering plan review process – also called an “architecture review” process by others. In the words of the report, the purpose of such a process is “to embody institutional and long-term interests in requirements for formulation, development, and post-deployment that might otherwise be compromised in favor of short-term goals.” Such engineering reviews would take place “at key points in the engineering process” and would be “conducted by peers and representatives of key stakeholders. These reviews are typically managed at the Program Executive Officer (PEO) level.” Within such a process, language is just one important engineering decision in software development. With advances in software engineering, it is now widely held that it does not make sense to focus solely on language selection and ignore other critical software-engineering decisions. A reasonable software-engineering process could work towards many of the same goals that the DOD had in adopting Ada. Considering proliferation of programming languages, for instance, the NRC report specifically addresses non-standard and proprietary languages. Under a systematic software-engineering process, it would be difficult to justify the use of non-standard, proprietary languages. Such a process, moreover, would enable Ada to be judged on its engineering merits rather than having a mandate intrude what Mr. Paige termed a “contentious point of resentment” in the software process. Timeframe At the NRC’s November 1996 public review of the Ada report, Dr. Barry Boehm, chair of the NRC study team, indicated that companies that have installed an engineering-review process find that it costs about one percent of development cost per project, and saves about ten percent. As valuable as the NRC report’s insights may be, software-engineering processes cannot be implemented overnight. Translating the NRC recommendations into some form of DOD-wide policy is a “non-trivial” task and will take some time to develop and implement. Both administratively and in terms of general policy guidance, it will require a disciplined analysis before implementation. Further, the DOD will have to consider the best way to implement the NRC report’s recommendation to make a significant investment in Ada infrastructure. It goes almost without saying that the direction of any such investment will be critical, and will require study and planning on its own. If you need more information Many DOD software project managers and developers have to think ahead in terms of years and decades. Such individuals may feel a need to know as much as possible about future decisions, and a desire to provide input to those decisions. To assist in this regard, there are Ada policy and waiver points of contact for each of the three Services and for the independent components (such as DISA, etc.). Contact the Ada Information Clearinghouse for information on finding suitable points of contact:
Ada Information Clearinghouse
|
