Lexical scoping, tail-recursion, the conceptual treatment of functions (as opposed to representations thereof) as data objects, and the ability to notate "anonymous" functions make SCHEME an excellent language in which to express program transformations and optimizations. Imperative constructs are easily modelled by applicative definitions. Anonymous functions make it easy to avoid needless duplication of code and conflict of variable names. A language with these properties is useful not only at the preliminary optimization level, but for expressing the results of decisions about order of evaluation and storage of temporary quantities. These properties make SCHEME as good a candidate as any for an UNCOL. The proper treatment of functions and function calls leads to generation of excellent imperative low-level code.
We have emphasized the ability to treat functions as data objects. we
should point out that one might want to have a very simple run-time environment
which did not support complex environment structures, or even stacks. Such an
end environment does not preclude the use of the techniques described here. Many
optimizations result in the elimination of LAMBDA-expressions; post CPS-conversion analysis eliminates the need to close many of the remaining LAMBDA-expressions. One could use the macros and internal representations of RABBIT to describe intermediate code transformations, and require that the final code not
actually create any closures. As a concrete example, imagine writing an
operating system in SCHEME, with machine words as the data domain (and functions
excluded from the run-time data domain). We could still meaningfully write, for
example:
