Exposing Behavioral Differences in Cross-Language API Mapping Relations

Abstract

Abstract. Due to various considerations, software vendors often translate their applications from one programming language to another, either manually or with the support of translation tools. Both these scenarios require translation of many call sites of API elements (i.e., classes, methods, and fields of API libraries). API mapping relations, either acquired by experienced programmers or already incorporated in translation tools, are much valuable in the translation process, since they describe mapping relations between source API elements and their equivalent target API elements. However, in an API mapping relation, a source API element and its target API elements may have behavioral differences, and such differences could lead to defects in the translated code. So far, to the best of our knowledge, there exists no previous study for exposing or understanding such differences. In this paper, we make the first attempt to expose and analyze behavioral differences in cross-language API mapping relations. From our result, we summarize eight findings and their implications that can improve effectiveness of translation tools, and also assist programmers in understanding the differences between mapped API elements of different languages during the translation process. Some exposed behavioral differences can indicate defects in translation tools, and four such new defects were confirmed by the developers of those tools.

Transferring an automated test generation tool to practice

“…However, Pex has been used by other third-party researchers from academia (who did not have direct collaboration with the Pex team on carrying out the academic research) in different ways. First, some researchers (e.g., [49][50][51][52]80]) made extensions of Pex by leveraging the APIs of Pex (without requiring to access the Pex source code). Second, some researchers (e.g., [33,37,56]) compared Pex (as a stand-alone state-of-the-art test generation tool) with their new approaches in empirical evaluations.…”

TestMig: migrating GUI test cases from iOS to Android

“…To support the conversion of system events, we need to construct their mappings between iOS and Android. The good part of system events is that, unlike GUI controls and events, they are predefined in the system, so we can construct a mapping in advance (although it may still need updates over time due to new iOS and Android versions) [54,61]. Furthermore, we need to record the system events being invoked when executing the iOS tests.…”

SAR: learning cross-language API mappings with little knowledge

“…Cross-Language Program Translation. For the problem of crosslanguage program translation, much work has utilized various statistical language models for tokens [23], phrases [15,24,25], or APIs [4,8,21,22,28,38,39]. A few studies also used word embedding for API mapping and migration (e.g., [12,13,26,28,35]), but our work does not need large number of manually specified parallel corpora or mapping seeds.…”