Gravitational-wave astronomy requires accurate predictions of radiation far beyond the boundaries of standard numerical simulations. A geometric framework to extract gravitational radiation is compactification at null infinity (scri), but this procedure introduces formally singular terms into the Einstein equations and are difficult to handle numerically. I will review the history of this problem and the approaches to deal with it in numerical relativity: the characteristic and hyperboloidal evolution problems. I will then present a numerical implementation combining Nützi’s recent regular nonlinear formulation combined with scri-fixing, which keeps null infinity at a fixed grid location. Demonstrations include nonlinear wave extraction from binary-black-hole mergers and matching to interior evolution on test problems.