Will they like this?: evaluating code contributions with language models

摘要: Popular open-source software projects receive and review contributions from a diverse array of developers, many whom have little to no prior involvement with the project. A recent survey reported that reviewers consider conformance project's code style be one top priorities when evaluating on Github. We propose quantitatively evaluate existence effects this phenomenon. To aim we use language models, which were shown accurately capture stylistic aspects code. find rejected change sets do contain significantly less similar project than accepted ones, furthermore, are more likely subject thorough review. Armed these results further investigate whether new contributors learn conform experience is positively correlated style.