Automated Data Structure Generation: Refuting Common Wisdom

Programming with enumerable sets of structures

“…This separation may significantly hamper the ability to control and reason about the generation process, which is necessary for performance [13]. Some tools eagerly explore a complete search space within the given bounds, even for enumeration of a few candidate instances.…”

“…Over a decade of research in automated testing has brought a variety of approaches that differ in the trade-off they make between expressiveness, efficiency, flexibility and ease of use [4,11,13,14,18,28,[42][43][44]. Our approach focuses primarily on providing what has been called bounded-exhaustive testing (BET), which amounts to generating all structures within a given bound [45] (in contrast to BET tools, however, it also supports enumeration from unbounded, infinite domains).…”

“…Different generation techniques provide different sets of features. While some techniques rely on a costly generate-and-test approach [22,32,42], efficient techniques usually avoid explicit enumeration of all possible structures [13,28,43]. Most of the prior work employs some form of backtracking: with a constraint solver [28,40,43], using custom search [4,40], or non-deterministic execution [18].…”

“…Often, the search problem is offloaded (after some translation) to a search engine, which means that the generator specifications highly dependent on the intricate details that affect the search. Although declarative approaches allow expressing constraints in a natural way, the performance of the generation process becomes highly sensitive to to the details of how the specification is written, sometimes unpredictably so [4,30,43]-seemingly equivalent specifications may differ drastically in performance [13], § 8.1. Consequently, in order to improve the impractical performance of generation, specifications become too complex and verbose, and may need to be tweaked, making them more complex and verbose [13,40]).…”

“…Although declarative approaches allow expressing constraints in a natural way, the performance of the generation process becomes highly sensitive to to the details of how the specification is written, sometimes unpredictably so [4,30,43]-seemingly equivalent specifications may differ drastically in performance [13], § 8.1. Consequently, in order to improve the impractical performance of generation, specifications become too complex and verbose, and may need to be tweaked, making them more complex and verbose [13,40]). The key advantages of declarative specification-namely concise expression of the desired properties and the ability to compose specifications in a modular fashion-may thus be lost.…”