An investigation was made of the characteristics of computer programming languages intended for the implementation of provably correct programs and of the characteristics of programs written in these languages. It was discovered that potential run time exceptions and the necessity of providing a rigorously correct implementation of exception handlers so dominate the potential control paths of programs written in verifiable languages that the usual code optimization techniques are ineffective. It was further discovered that the call intensive control structures of these programs, necessitated by verification constraints, also thwart optimization and lead to inefficient code. It is shown that theorems can be derived at potential exception sites which, if true, guarantee that the exception condition will never arise permitting removal of the exception path from the program’s flow graph. These theorems are proved using the automatic theorem prover which is part of the program verification system. Is is also shown that many of the routine calls contained in verifiable programs may be reduced in expense by converting parameters to global variables or eliminated completely by expanding the called routines at their call sites. Both the exception suppression and call reduction techniques reduce the complexity of the program’s call graph and facilitate conventional optimizations. Several examples are presented and the potential improvements in code size resulting from the application of these techniques are discussed.

Nearly all algorithms for code optimization use a control flow graph. In this thesis, I will show that with very minor restrictions on program structure, an abstract syntax tree can be used instead, leading to algorithms that are often much simpler than their graphbased counterparts. The conclusion is that abstract syntax trees, not control flow graphs, should be the fundamental data structure in code optimization. 1 Introduction Most optimizing compilers consist of a front-end that does syntactic and semantic analysis, and a back-end that does optimization and machine code generation [ASU86]. The main data structure in the front-end is an abstract syntax tree (AST), while in the back-end it is a control flow graph (CFG), which consists of nodes representing computations and edges representing control flow. Thus, code optimization operates on a data structure---the CFG---in which the program has essentially been "flattened" into a tangle of GOTOs (edges). Unlike an AST, a CFG c...