Code completion is a widely used feature of modern integrated development environments. This study examines the ways in which code completion is used by professional software developers, as well as their actions when code completion doesn't offer the expected results. We observe that code completion is used with the intention of writing code faster and more correctly, that a large fraction of the code completions are not accepted by the users and that users often used code completion as a debugging tool: when the suggestions are not useful or not expected they are seen as a indicator that there is an error in the program.
Type of Publication: Paper
Conference: PPIG 2015 - 26th Annual Workshop
Publication Year: 2015
Paper #: 14