NASA Logo

NTRS

NTRS - NASA Technical Reports Server

Press Enter or click the Search button to begin your search.

Back to Results
Traceability Through Automatic Program GenerationProgram synthesis is a technique for automatically deriving programs from specifications of their behavior. One of the arguments made in favour of program synthesis is that it allows one to trace from the specification to the program. One way in which traceability information can be derived is to augment the program synthesis system so that manipulations and calculations it carries out during the synthesis process are annotated with information on what the manipulations and calculations were and why they were made. This information is then accumulated throughout the synthesis process, at the end of which, every artifact produced by the synthesis is annotated with a complete history relating it to every other artifact (including the source specification) which influenced its construction. This approach requires modification of the entire synthesis system - which is labor-intensive and hard to do without influencing its behavior. In this paper, we introduce a novel, lightweight technique for deriving traceability from a program specification to the corresponding synthesized code. Once a program has been successfully synthesized from a specification, small changes are systematically made to the specification and the effects on the synthesized program observed. We have partially automated the technique and applied it in an experiment to one of our program synthesis systems, AUTOFILTER, and to the GNU C compiler, GCC. The results are promising: 1. Manual inspection of the results indicates that most of the connections derived from the source (a specification in the case of AUTOFILTER, C source code in the case of GCC) to its generated target (C source code in the case of AUTOFILTER, assembly language code in the case of GCC) are correct. 2. Around half of the lines in the target can be traced to at least one line of the source. 3. Small changes in the source often induce only small changes in the target.
Document ID
20030107320
Acquisition Source
Ames Research Center
Document Type
Preprint (Draft being sent to journal)
Authors
Richardson, Julian
(Research Inst. for Advanced Computer Science Moffett Field, CA, United States)
Green, Jeff
(QSS Group, Inc. Moffett Field, CA, United States)
Date Acquired
September 7, 2013
Publication Date
January 1, 2003
Subject Category
Cybernetics, Artificial Intelligence And Robotics
Meeting Information
Meeting: 2nd International Workshop on Traceability in Emerging Forms of Software Engineering
Location: Montreal
Country: Canada
Start Date: October 6, 2003
End Date: October 10, 2003
Distribution Limits
Public
Copyright
Work of the US Gov. Public Use Permitted.
No Preview Available