Sharlit—a tool for building optimizers

• Authors: Steven W. K. Tjiang;John L. Hennessy
• Venue: PLDI '92 Proceedings of the ACM SIGPLAN 1992 conference on Programming language design and implementation
• Year: 1992

A complex and time-consuming function of a modern compiler is global optimization. Unlike other functions of a compiler such as parsing and code generation which examine only one statement or one basic block at a time, optimizers are much larger in scope, examining and changing large portions of a program all at once. The larger scope means optimizers must perform many program transformations. Each of these transformations makes its own particular demands on the internal representation of programs; each can interact with and depend on the others in different ways. This makes optimizers large and complex.Despite their complexity, few tools exist to help in building optimizers. This is in stark contrast with other parts of the compiler where years of experience have culminated in tools with which these other parts can be constructed easily. For example, parser generators are used to build front-ends, and peephole optimizers and tree matchers are used to build code generators.This paper presents Sharlit, a tool to support the construction of modular and extensible global optimizers. We will show how Sharlit helps in constructing data-flow analyzers and the transformations that use data-flow analysis information, both are major components of any optimizer.Sharlit is implemented in C++ and uses C++ in the same way that YACC uses C. Thus we assume the reader has some familiarity with C++[9].