Back to FAQ
12 July 1996
For Presentation to the National Research Council
The AIA. The Aerospace Industries Association (AIA) is a trade association that represents most of the aerospace companies in the United States. These companies are manufacturers of commercial, military, and business aircraft, helicopters, aircraft engines, missiles, spacecraft, and related equipment and components.
The National Research Council Study of Ada within DOD. Earlier this year, the Computer Science and Telecommunications Board (CSTB) of the National Research Council (NRC) formed a study group to review "the past and present contexts for using Ada within the Department of Defense (DOD)". The NRC study group invited AIA to provide an industry position on Ada. This document, AIA Position Statement on Ada, was developed in response to the NRC request. It includes AIA recommendations, which are followed by a review of the basis for those recommendations.
At a time when extraordinary steps are being taken to encourage commercial solutions and large-scale reuse, program management on both the government and industry sides should have more freedom to determine what best meets the life-cycle cost objectives of the project. We advise the DOD to change its computer language policy now and end the exclusive mandate for Ada for all programs, including those for which the government maintains the software.
The AIA believes that a decision regarding which computer language or languages to use on a given DOD project is an implementation issue that should be based on a trade study. The study, which should be conducted within the context of a specific project or a specific family of projects, should include considerations of performance-based specifications, interoperability, reuse of an existing system (if any), open systems, use of commercial off-the-shelf (COTS) items, tool environment quality, cost and availability of tools, configuration management, and life-cycle considerations. Such a study should be part of the requirements definition process; there should be no presumption in favor of any particular language. The study could be conducted either prior to contract award or after the contract has been awarded. In either case, an integrated team including the developer (or potential developers), the supporters, the maintainers, and the testers all should advise on the language decision. If the study were conducted prior to contract award, the study results could be included in the RFP. The trade study should document the decision, the rationale, and the decision process, and should be retained in the government program office.
In a joint effort (lasting at most six months) DOD and industry should develop guidance on computer language trade studies. An experienced, high-level, joint government and industry team of "stakeholders" (that is, parties who for many years have been in the business of developing, acquiring, or supporting DOD systems) should be tasked to develop this guidance. This team, which should contain both systems and software expertise, should work to ensure that DOD life-cycle considerations are acknowledged and satisfied. They should give consideration to how certain kinds of systems may merit special consideration; e.g., safety/flight critical systems, or systems where the size of the software exceeds a threshold. Other factors include whether software is embedded, potential future modernizations, and interfaces with other systems and software. The guidance should deal with the unfortunate fact that some managers on both the government and industry side may have insufficient knowledge of Ada capabilities; under pressure such people might ignore life-cycle concerns, or might turn away from essential up-front design analysis. AIA believes that life-cycle cost studies and up-front design analysis should precede language selection.
AIA strongly supports the use of Ada in systems where its use makes both engineering and economic sense. Our position that the exclusive mandate for Ada should end is not an attack on the value of Ada. Indeed, we believe that Ada no longer needs a mandate in part because of the proven strengths of Ada 83 and the improvements that have been made in Ada 95. AIA believes that adopting the recommendations presented herein is an essential first step in moving toward a broad consensus position on computer languages within the defense industrial community, in appropriately responding to the thrust of the SECDEF memorandum regarding performance specifications and commercial standards, and in providing due consideration to reducing costs through the increased use of commercial products and practices.
Basis for AIA Recommendations
The Importance of Software. Software is a critical and increasingly significant factor in the systems and subsystems that are created by our member companies. In many cases the software represents the integration technology for the system or subsystem as a whole. For long-life weapon systems a software language is a long-enduring element.
The Merits of Ada. The implementation of the Ada programming language within the DOD has been a significant factor in the way AIA companies deal with issues of large-scale software development and systems integration. Ada 83 was designed to support sound software engineering, especially in large systems with a long life-cycle; noteworthy features of ADA 83 include strong typing and an inherent reliability based on extensive error detection at both compile-time and run-time. Ada 95 continues that tradition with increased support for potential software reuse and for object-oriented programming, and increased flexibility for real-time systems programming. Additionally, Ada has an ability that is unique among high order languages (HOL) to be a design representation language; and it has been shown that in large systems the defect rate for Ada is significantly less than that for the more popular commercial HOLs. We make these observations not to "damn with faint praise"; rather we wish to emphasize that Ada has significant technical merits.
The Limited Acceptance of Ada. The acceptance of Ada has not been as widespread as its proponents have wished of it. In the first place, Ada has only limited acceptance in the US commercial world. There are many languages with a wider commercial acceptance than Ada, such as Basic, C, COBOL, and FORTRAN; and there are newer languages that are gaining in commercial popularity, such as C++ and Java. In addition, Ada has some but only limited acceptance in the US academic world. Most researchers in universities create small computer programs for which Ada has fewer advantages. According to recent data from the DOD Ada Joint Program Office (AJPO), over 20% of colleges and universities offer courses on Ada. Many Ada proponents believe that this percentage would be greater but for the lack of inexpensive Ada compilers and tools. It remains to be seen whether the new availability of free Ada 95 compilers for a wide variety of platforms (including personal computers) and other promotion efforts will result in a greater use of Ada in the academic world and in commerce. At the present time Ada continues to be encumbered by poor exposure in the US commercial world.
Ada and Development Costs. Because of these acceptance issues, the development costs (as contrasted with life-cycle costs) associated with Ada become an issue that may impact competitiveness. There is an ongoing cost to industry and to government software development groups for Ada training, since in many areas of the country the schools do not provide adequate courses. The cost of industrial-strength Ada language compilers and other support tools is high relative to other languages that have a broader commercial base. Availability issues may drive up development costs, also. The use of commercial hardware platforms for embedded systems is increasing, and the frequency of performance enhancements in those platforms for both host and target systems has increased. At the same time, because of the support for a commercial language environment by commercial hardware vendors, Ada support often lags. This may result in less use of Ada, as commercial technology development supplants customized development.
AIA Experiences with Ada Vary. AIA member companies have experienced a wide range of results from the use of Ada 83. All have greatly benefited from the software engineering support features of Ada, including reduced error rates. Many can cite projects in which Ada has had a substantial positive impact, overall. Of special note in this regard are large, highly visible projects such as F-22, BSY-2, Boeing 777, and Peace Shield. However, some have suffered because Ada support tools were not robust nor available when needed, or because Ada presented interface difficulties in heterogeneous environments.
The Ada Mandate and Performance Standards. Current policy in DOD on computer languages consists of an exclusive mandate for the use of Ada across the DOD. Two years ago, on June 29, 1994, a SECDEF Memorandum, "Specifications and Standards - a New Way of Doing Business," was issued. This memorandum directs the use of performance standards, more reliance on commercial standards as a priority and greater dependence on the commercial industrial base. On August 26, 1994, E. Paige and N. Longuemare issued a follow-on memorandum concerning the use of Ada, reiterating the DOD commitment to Ada, and noting that the Ada mandate does not conflict with the SECDEF Memorandum. Most recently this was reiterated in the new 5000 series guidance. Throughout DOD there is a need for more affordable weapon systems that use open architecture approaches to combat life-cycle cost growth during modernization. In this context AIA believes that there still are open issues regarding Ada software and the desire to integrate the defense and commercial industrial base.
Acquisition Reform and Ada. The DOD is moving toward permitting an increased reliance on commercial products, practices, and processes. Recently, acquisition reform initiatives have supported this trend; there is a general movement toward having government specifications focus on performance requirements - "what" is required from a system, as opposed to "how" to design or implement the system. AIA supports these reforms and believes that usually the selection of a programming language is an implementation detail rather than a performance requirement. As such the selection of a computer language should be based on both business judgment and engineering judgment prior to or during development, with no initial presumption in favor of any specific choice. Frequently the choice has clear life-cycle implications, and so should be made with input from the support and maintenance organizations (if they are different from the developers).
Limiting the Number of Programming Languages. AIA recognizes that DOD has an interest in not having to support systems in hundreds of different programming languages. In the 1970s and 1980s it may have seemed necessary to mandate a single language to force a reduction in support costs. In the 1990s and beyond, AIA believes that the commercial marketplace will of its own accord drive toward a relatively small number of commercially viable languages. Going from hundreds of languages to several is virtually the same as going from hundreds to one, in terms of the benefit to the DOD. In addition, we have many reports that the current waiver process, which is used in the attempt to enforce the single language policy, is costly and disruptive; a more open policy would reduce the cost and disruption.
Life-Cycle Cost Issues. Certainly the DOD has an interest in encouraging its acquisition program managers to promote engineering solutions that address life-cycle issues. These issues include life-cycle cost, supportability, maintainability, and adaptability, all of which were considerations in the design of Ada. However, technology changes with time, and an equally important life-cycle consideration is the level of support for a language during that life-cycle. In the US, Ada continues to be primarily a niche language for the DOD. AIA believes that there is nothing wrong with efforts to promote Ada, to make Ada technology more easily available to faculty and students as well as to create a wider commercial base for Ada. Only with wider commercial acceptance of Ada and its support and promotion by a major vendor (e.g., IBM, HP), is it likely that all the anticipated life-cycle savings will be achieved. Ada should be maintained for systems when it makes business and technical sense. At the same time we should provide for a transition to newer technologies that may have a greater impact on reducing life-cycle costs.
Fears from Eliminating and Fears from Keeping the Mandate. Within both industry and government, there are some who believe that the exclusive mandate for Ada is essential if Ada is to build on its successes to date and continue to succeed in the future. Such proponents may equate ending the exclusive mandate with "killing" Ada. They fear that without such a mandate vendors may lessen their support for Ada products; also, they fear that without the mandate projects would not be forced to address software life-cycle cost issues. Such concerns are not groundless; it is hard to predict the future with total confidence. However, we believe that the "doomsayers" are wrong. Ada is too good a language for large embedded systems; the marketplace would not let it die. AIA member companies have made substantial investments in Ada technology, and have a clear appreciation of its value. In the future as in the past, with or without a mandate, Ada will be a strong contender for large real-time systems that require ongoing maintenance and support (e.g., the Joint Strike Fighter (JSF) Program). It may be a strong contender for other kinds of systems, as well. The language has proven its value; after the recent improvements in Ada 95, the language is ready to stand on its own. AIA has a concern that to continue the exclusive mandate for Ada could inhibit the use of commercial best practices, prevent dual-use applications, harm American competitiveness, and inhibit life-cycle benefits that could result from the judicious adoption of commercial products and practices.