In this paper, we present a new system for proof-search in propositional intuitionistic logic from which an efficient implementation based on structural sharing is naturally derived. The way to solve the problem of formula duplication is not based on logical solutions but on an appropriate representation of sequents with a direct impact on sharing and therefore on the implementation. Then, the proof-search is based on a finer control of the resources and has a ${\cal O}(n\log n)$-space complexity. The system has the subformula property and leads to an algorithm that, for a given sequent, constructs a proof or generates a counter-model.