Abstract
The boundary construction for spacetimes suggested by Geroch, Kronheimer and Penrose in the early seventies presents some important obstacles when both, the past and the future boundaries are considered simultaneously. Since then, different authors have tried to solve these questions with new approaches to the causal boundary, without totally satisfactory results. In this talk I suggest a new development of the GKP construction intended to provide a solution to these problems. We will analyze the properties of this new approach, compare it with previous constructions and compute it in some physical examples.