Towards Trustworthy Refactoring in Erlang

Abstract

Tool-assisted refactoring transformations must be trustworthy if programmers are to be confident in applying them on arbitrarily extensive and complex code in order to improve style or efficiency. We propose a simple, high-level but rigorous, notation for defining refactoring transformations in Erlang, and show that this notation provides an extensible, verifiable and executable specification language for refactoring. To demonstrate the applicability of our approach, we show how to define and verify a number of example refactorings in the system.

Trustworthy Refactoring via Decomposition and Schemes: A Complex Case Study

“…A special case of function refactoring is function signature refactoring, which only transforms the head of the function definition and its refereces. As we demonstrated in our previous paper [8], this scheme can be used as well for renaming a function and to restructure or reorder its arguments.…”

“…In our prior work [8], we introduced novel abstractions for executable and verifiable refactoring definitions. Besides incorporating the basic idea of decomposition to micro-refactoring, we use a semantic program graph representation to define transformations in terms of strategic term rewriting rules extended with language-level semantic conditions.…”

“…We express refactoring correctness in terms of a set of equivalence formulas. In our previous work [8] we presented how to use a proof system to prove refactorings whose correctness can be expressed by the equivalence of two expression patterns under a given condition. We reduced the equivalence property of the two expression patterns to a correctness property of an aggregated program constructed by the two expression patterns (according to [4]), then we applied the language-independent, general-purpose proof system to automatically check the validity of our property.…”

“…For the syntax and semantics of the refactoring formalism, we refer to the comprehensive introduction presented in [8]. Roughly summarising, composite functions define a refactoring sequence, whilst prime functions instantiate schemes with rewrite rules applied to the definition and the references of some data or name.…”

“…The general-purpose reachability proof system, referred to in our previous publication [8], can generate proofs automatically, but it can only prove partial equivalence and not full equivalence. Moreover, this generic proof system generates unnecessarily many branches, because it cannot take into account the specialities of equivalence proofs of deterministic programs.…”