Conditional upgrade and installation of software based on risk-based validation

Abstract

Embodiments are directed to reducing the number of unit tests performed within a system. In one scenario, a computer system accesses a specified version of an application and evaluates that version of the application to identify connections between software features in the application, and to further identify defects in the software features. The computer system also determines a risk level for the software features, which indicates a likelihood of that software feature causing a software malfunction according to the identified connections or defects. The computer system further generates a feature list that includes, for at least some of the features, a corresponding ongoing risk indicator that indicates the determined likelihood of that feature causing the defect or software malfunction, and performs unit tests against the software application for those features that were indicated as sufficiently likely to cause a defect or software malfunction upon installation or upgrade of the application.