The aim of this work in progress is twofold: first to find out a significant, as large as possible subset of the Fluid Construction Grammar (FCG) formalism that can be supported by an efficient implementation, and second to check whether the LIGHT platform which until now was used for running large scale unification grammars can be (and if so, how it will be) extended so as to support evolving grammars like those defined in the FCG framework. This paper justify and documents some of the basic principles that we defined and partially implemented for translating FCG into LIGHT.