Detection of Non-Literal Errors Through the Integration of Manually and Automatically Generated Detection Rules

Thesis Abstract:
A non-literal error is defined as an error message given by the compiler that doesn’t correspond to the actual error present in the code. In these cases, debugging code can be a difficult even for seasoned programmers. Continuing recent work on most frequently committed non-literal errors, we plan to improve on the current detection system by integrating manually and automatically generated detection rules into the already existing error detection system. We believe this will enhance the capability of the detector by improving the methods for which it scans code for errors.
Members:
Joshua Bautista
Jaime Anson