Random generation of well-typed terms lies at the core of effective random testing of compilers for functional languages. Existing techniques have had success following a top-down type-oriented approach to generation that makes choices locally, which suffers from an inherent limitation: the type of an expression is often generated independently from the expression itself. Such generation frequently yields functions with argument types that cannot be used to produce a result in a meaningful way, leaving those arguments unused. Such “use-less” functions can hinder both performance, as the argument generation code is dead but still needs to be compiled, and effectiveness, as a lot of interesting optimizations are tested less frequently. In this paper, we introduce a novel algorithm that is significantly more effective at generating functions that use their arguments. We formalize both the “local” and the “nonlocal” algorithms as step-relations in an extension of the simply-typed lambda calculus with type and arguments holes, showing how delaying the generation of types for subexpressions by allowing nonlocal generation steps leads to “useful” functions. We implement our algorithm demonstrating that it’s much closer to real programs in terms of argument usage rate, and we replicate a case study from the literature that finds bugs in the strictness analyzer of GHC, with our approach finding bugs four times faster than the current state-of-the-art local approach.

Enumerations represented as bijections between the natural numbers and elements of some given type have recently garnered interest in property-based testing because of their efficiency and flexibility. There are, however, many ways of defining these bijections, some of which are better than others. This paper offers a new property of enumeration combinators called fairness that identifies enumeration combinators that are better suited to property-based testing. Intuitively, the result of a fair combinator indexes into its argument enumerations equally when constructing its result. For example, extracting the nth element from our enumeration of three-tuples indexes about $\sqrt[3]{𝑛}$ elements into each of its components instead of, say, indexing $\sqrt[2]{𝑛}$ into one and $\sqrt[4]{𝑛}$ into the other two, as you would if a three-tuple were built out of nested pairs. Similarly, extracting the nth element from our enumeration of a three-way union returns an element that is $\frac{𝑛}{3}$ into one of the argument enumerators. The paper presents a semantics of enumeration combinators, a theory of fairness, proofs establishing fairness of our new combinators and that some combinations of fair combinators are not fair. We also report on an evaluation of fairness for the purpose of finding bugs in programming-language models. We show that fair enumeration combinators have complementary strengths to an existing, well-tuned ad hoc random generator (better on short time scales and worse on long time scales) and that using unfair combinators is worse across the board.