WIP: Live Restructuring of Data Architecture

In large-scale Earth System simulation codes, such asthe Accelerated Climate Model for Energy (ACME), complex user derived data types (containing large numberof variables) are designed to represent the interactionsof atmosphere, ocean, land, ice, and biosphere toproject global climate under a wide variety of conditions. The following is our proposed approach to restructurethe data architecture of a land component within theACME project while the project is undergoing activedevelopment. The data architect for the land subsystemdefines the new datatype requirements that wouldgreatly simplify the implementation of terrestrial landsubmodels by converting more than 50 to just eight primarydata-types. Since the code is developed with thecommunity governance, we have to ensure that the restructuringdoes not interface the other developmentwhich, with dozens of changes occurring every day, makeit impossible to work on a shared development branch. The active development also occurs on almost five hundredbranches, making it extremely difficult to assesspotential interactions. To address these challenges we have designed andstarted an iterative procedure for implementing the datarestructuring and estimating both the effort it takes torestructure and the effort would save once the restructuringis implemented.