The paper introduces BBS (Branch-and-Bound Skyline), a progressive algorithm for skyline computation, which is based on nearest neighbor search. Unlike the current most efficient algorithm, NN, BBS is IO optimal, meaning it only accesses R-tree nodes that may contain skyline points and does not retrieve duplicates. BBS also incurs significantly less space overhead compared to NN. The algorithm is simple to implement and can be applied to various variations of skyline queries, such as ranked, constrained, dynamic, and $K$-dominating queries. Experimental results show that BBS outperforms NN in terms of CPU and IO costs for all problem instances, often by orders of magnitude. The paper also discusses the limitations of NN, including the need for duplicate elimination, multiple node visits, and large space requirements, and provides a detailed analysis of BBS's correctness, optimality, and efficiency.The paper introduces BBS (Branch-and-Bound Skyline), a progressive algorithm for skyline computation, which is based on nearest neighbor search. Unlike the current most efficient algorithm, NN, BBS is IO optimal, meaning it only accesses R-tree nodes that may contain skyline points and does not retrieve duplicates. BBS also incurs significantly less space overhead compared to NN. The algorithm is simple to implement and can be applied to various variations of skyline queries, such as ranked, constrained, dynamic, and $K$-dominating queries. Experimental results show that BBS outperforms NN in terms of CPU and IO costs for all problem instances, often by orders of magnitude. The paper also discusses the limitations of NN, including the need for duplicate elimination, multiple node visits, and large space requirements, and provides a detailed analysis of BBS's correctness, optimality, and efficiency.