Is Our Merging Right? Towards More Reliable Merging Decision

In the distributed and collaborative development environment, where a optimistic version controlling is used, software developer inevitably have to manually resolve the conflict while merging. There are a number of tools (Git, SVN, CVS, AMOR, IBM Jazz, etc.) that support automatic merging for two non-conflicting segment of two versions of a program or a model. However, these tools have very little to offer for resolving the conflict of the programs or the models, and the conflict resolution is largely dependent on the manual intervention. Often, a developer may not have sufficient knowledge about the conflicting blocks that he is merging, and then he makes guesses based on his experience and some knowledge on the problem domain. These illiterate guesses, if wrong, may cause unnecessary reworks: identify the merging problem, consult with team-mates to resolve the conflict correctly, redeploy, re-regression, testing, etc. The consequence is more catastrophic if the merging is wrong and that is not identified before deployment. In this project, we explore the state of the art conflict resolution tools for model versioning and identify a problem, outline a model of provenance information semantic, and propose a framework to potentially help the developer to take better merging decisions.