Abstract:Bug fix is an important and challenging task in software development and maintenance. Bug fix is also a dangerous change, because it might induce new bugs. It is difficult to decide whether a bug fix is safe in practice. In this paper, we conducted an empirical study on bug inducing analysis to discover the types and features of fault prone bug fixes. We use a classical algorithm to track the location of the code changes introducing the bugs. The change types of the codes will be checked by an automatic tool and whether this change is a bug fix change is recorded. We analyze the statistics to find out what types of change are most prone to induce new bugs when they are intended to fix a bug. Finally, some guidelines are provided to help developers prevent such fault prone bug fixes.

Bug Inducing Analysis to Prevent Fault Prone Bug Fixes.