Even though the satellite proposal was not successful, the experiment demonstrated the utility of our approach and gave the lab some credibility. The aerospace engineer that participated in the experiment became a very strong advocate for the technology. By necessity (e.g., lack of time), we had created a solution that was simple and fit into the existing development process with minimal impact. As a result the initial crude prototype was further developed into a usable system, the Flight Code Generator (FCG), that is actively used on several programs. The current version of the system employs dataflow analysis, various code optimization techniques, type inference, and analysis of finite state machines.
FCG is successful because it (i) is specialized to a narrow domain, (ii) generates code that fits into an existing architecture, and (iii) fits into an established development process. The following is a brief description of these technical aspects of FCG.