It will be demonstrated, how theory helps to provide a good background to actual realization of computations in the field of Gabor analysis. Originally thought to be highly computational complex and instable (due to the non-orthogonality of the Gabor system) one knows now much better how to handle issues of redunadancy (i.e. linear dependencies of Gabor families) and numerical stability. Function space methods help to connect the continuous model with the finite dimensional approximation realized (approximately) on the computer. If time permits the role of the Wexler-Raz (or Ron-Shen) duality principle for mobile communication can be addressed.