Report suggests DOD limit Ada use

The National Research Council has recommended that the Defense Department roll back its longstanding often unpopular mandate requiring the Ada programming language and reserve it for use only in weapons systems and other warfighting software.

The recommendations are part of "Ada and Beyond: Software Policies for the Department of Defense " a report prepared by NRC's Computer Science and Telecommunications Board. Emmett Paige Jr. assistant secretary of Defense for command control communications and intelligence requested that the NRC board review DOD's policy which identifies Ada as the language of choice for major software development or re-engineering efforts. DOD maintains about 50 million lines of code.

The current mandate covers major automated information systems programs in areas such as asset monitoring logistics and office and management support. That would not be the case in the mandate proposed by NRC.

"In commercially dominated areas using Ada is generally less cost-effective than using other languages " the board decided. "Requiring Ada's use in commercially dominated applications would place DOD systems at a competitive disadvantage."

The board also advised DOD to invest some $15 million a year in Ada "infrastructure support" - essentially seed money - and dismantle its Ada waiver process in favor of a wider Software Engineering Plan Review process.

Paige who was briefed along with other DOD officials Oct. 31 said in a statement "We will be looking closely at how we can implement the intent and spirit of the National Research Council's recommendations and will have more to say after the completion of that look."

The policy in question stems from DOD's longstanding interest in having a language robust enough to run in real-time high-availability environments. DOD helped spawn the market for Ada in the late 1970s and early 1980s because it did not believe any existing languages were sufficient.

The market has changed considerably in recent years with the emergence of C C++ and other languages in the commercial environment but Ada still is seen by many observers as the best choice for large-scale mission-critical programs.

The NRC committee proceeded on that same assumption of Ada's advantages for that given class of applications but interprets that class in a different manner than DOD has in many cases to date.

To make clear the distinction between its recommendations and the current policy the committee used the term "warfighting" rather than "command and control " said chairman Barry Boehm a professor at the University of Southern California and a former director of the Information Sciences and Technology Office at the Defense Advanced Research Projects Agency. Other applications that often fall under the heading of "command and control" - such as many logistics and other related support operations - do not meet the same criteria Boehm said. "These are things where your best solutions are gluing [commercial off-the-shelf] products together " he said. "Our bottom line is that Ada does not give you an advantage there."

For other applications the committee favors considering Ada as part of a Software Engineering Plan Review that considers language choice in conjunction with factors such as system architecture life-cycle costs and other issues. DOD should establish review councils to consider such decisions the panel said rather than having a waiver process that focuses strictly on programming language.

The interpretation of "warfighting software" is likely to meet resistance among Ada proponents in DOD industry observers said Paige and others favor the larger "command and control" umbrella.In his statement Paige said DOD officials "appreciate the council's recognition that the Department of Defense continues to need policy in this area and their recommendation that Ada be maintained as the programming language for warfighting systems."

However he added "We are certain that we have to better define the term `warfighting systems.'