We are currently capturing (reverse engineering and documenting) a representation of the module structure of the Hadoop Distributed File System (HDFS), version 0.21. We are doing this by analyzing the HDFS source code, specifically the dependencies in the code. By looking for modularity violations using this representation, we are able to suggest a number of improvements to the modularity of HDFS.
We would like input from you, contributers of HDFS, on the correctness of the refactoring possibilities that we have identified, and on the feasibility of actually making the proposed changes. In return, you are getting documentation of the module structure of the HDFS, and a possibly useful list of changes that could be made to improve the maintainability of the source code.