Supporting source code difference analysis

The paper describes an approach to easily conduct analysis of source-code differences. The approach is termed meta-differencing to reflect the fact that additional knowledge of the differences can be automatically derived. Meta-differencing is supported by an underlying source-code representation developed by the authors. The representation, srcML, is an XML format that explicitly embeds abstract syntax within the source code while preserving the documentary structure as dictated by the developer. XML tools are leveraged together with standard differencing utilities (i.e., diff,) to generate a meta-difference. The meta-difference is also represented in an XML format called srcDiff. The meta-difference contains specific syntactic information regarding the source-code changes. In turn this can be queried and searched with XML tools for the purpose of extracting information about the specifics of the changes. A case study of using the meta-differencing approach on an open-source system is presented to demonstrate its usefulness and validity.