The development of applications that target dynamic networks often adresses the same difficulties. Since the underlying network topology is unstable, the application has to handle it in a context-reactive manner, and, when possible, with algorithms that are localized and decentralized. The present paper shows how the DA-GRS model, that is a high level abstract model for localized graph algorithms, can be used to drastically improve the development time of selforganized systems. The canonical method that is proposed makes use of generic pieces of code, along with some additional code generation. All the concepts presented here are illustrated by means of a spanning forest algorithm, that can then be used by the application at a high abstraction level.