Prioritizing lingering bugs

Abstract

As the software projects become more complex, the release deci-sion is made without resolving all the bugs in the issue tracking system. Accumulation of the bugs in the bug repository is similar to nancial obligation as we borrow time and resources to engage in another activity rather than resolving the bugs. Deferring the bug in the next release may have some consequences. Therefore, the decision whether to resolve the bug in the current release or postponing it to the next release is a crucial decision. In this proposal, we study the deferred bugs (lingering bugs) against the non-deferred bugs (regular bugs). Our aim is to develop the pre-dictive model which can predict whether the bug would linger or not. Additionally, we are interested in measuring of the linger-ing bug in terms of principal (standard time it takes to x them) and risk of liability (impact). We propose to use reinforcement learning for prioritization of lingering bugs with respect to their impact.