An undo operation is executed by an application by performing the inverse actions of the do operation to which the undo operation relates. Previous designs simply swapped memory to execute an undo operation according to code that was entirely dissimilar to the code of the do operation. The dissimilarity of the code caused debugging such operations to be difficult. Using the inverse action to undo an action results in the similar code for the do, undo, and redo operations. Also, undo atoms are logged for do, undo, and redo operations so that any exceptions thrown during an operation allows the application to return to a previous, consistent state by operating on the undo atoms.