Complete completion using types and weights

Abstract

Developing modern software typically involves composing functionality from existing libraries. This task is difficult because libraries may expose many methods to the developer. To help developers in such scenarios, we present a technique that synthesizes and suggests valid expressions of a given type at a given program point. As the basis of our technique we use type inhabitation for lambda calculus terms in long normal form. We introduce a succinct representation for type judgements that merges types into equivalence classes to reduce the search space, then reconstructs any desired number of solutions on demand. Furthermore, we introduce a method to rank solutions based on weights derived from a corpus of code. We implemented the algorithm and deployed it as a plugin for the Eclipse IDE for Scala. We show that the techniques we incorporated greatly increase the effectiveness of the approach. Our evaluation benchmarks are code examples from programming practice; we make them available for future comparisons.

Automatically Generating Test Templates from Test Names (N)

“…Various code completion techniques have been developed based on Hidden Markov models (HMMs), ranking, statistical properties of source code, machine learning, data mining, as well as other matching approaches to recommend APIs according to users' incomplete inputs (e.g., [4,6]). …”

Towards standardized evaluation of developer-assistance tools

“…Code recommenders are assistance tools that propose different source-code elements, like method calls [3,6,18], method-call parameters [24], method-call chains [5,14], or whole example-code snippets [7,15,17,20,21]. In the remainder of this work we use the more general term "recommenders" to refer to them.…”

Refactoring with synthesis

“…10 For one of the tasks (Task 3), RESYNTH cannot, in fact, find the desired solution, but comes up with a slightly different solution.…”

“…PARSEWeb [34] addresses the same problem, but its search is guided by existing source code mined from the web, which helps eliminate many undesirable API sequences. More recent work [10,23] searches for suitable expressions of a given type at a particular program, considering more of the program context around that point. These systems rely on good ranking algorithms to handle large numbers of potentially-suitable expressions.…”