"A complete rewrite of legacy science software, with a specific focus on clarity, can prove useful. It uncovers bugs, provide a definitive statement of the underlying algorithm, lets scientists easily pose and answer new research questions, makes it easier to develop new visualizations and other interfaces, and makes the code considerably more accessible to interested third parties and the public."

[N. Barnes and D. Jones, 2011: Clear Climate Code: Rewriting Legacy Science Software for Clarity. IEEE Software, Nov/Dec 2011, 36 - 42.]

The YAC development started as a pilot study to investigate the coupling between unstructured meshes which gradually turned into a mature fully parallelised coupling software. Rather than implementing all required functionality directly into an existing coupler like OASIS4 we decided to ground our effort on a light weight software infrastructure. YAC2 is a redesign of YAC1. All interpolation methods available in YAC1 are supported with YAC2. In contrast to YAC1 halo information is no longr required. In order to keep the software light we still stick to:

Only unstructured and fully regular grids are supported directly.
Internally, grid information and exchange fields are handled as double.
Fortran reals are converted to double precision before delivering it to the C world.