Specialization of SARA for software synthesis

The SARA system has been developed to permit top-down design without necessarily prespecifying that implementation would be done in software or hardware. It was felt very important to give the designer freedom to make that decision at a late stage and therefore select the appropriate place for the hardware-software interface. However, the synthesis methodology does not force the designer to hide an intention to use particular devices or particular software systems. Moreover, the area in greatest need of design assistance is programming where a language and associated virtual or real machine would already have been selected. Hence the question was raised, “What specialization of SARA would be important for the synthesis of software systems?” This paper discusses models of software systems under two very different conditions. In one, it is known that the design of a program for a sequential PL/I processor raises no synchronization issues to be managed by explicit flow-of-control models. Using a simple classical example, the paper shows that it is possible to describe a multi-level model, using SARA, which goes smoothly from programming-in-the-large to programming-in-the-small or actual code. In a second simple case synchronization issues are vital and it is important to keep a token machine explicitly in the model.