Oxygen is a much better evolutionary index than iron to describe the history of lithium-beryllium-boron (LiBeB) since it is the main producer of these light elements at least in the early Galaxy. The O-Fe relation is crucial to the determination of the exact physical process responsible for the LiBeB production. At low metallicity, if [O/Fe] vs. [Fe/H] is flat, then the production mode is independent of the interstellar metallicity, BeB is proportional to oxygen, i.e. is of primary nature. If not, the production mode is a function of the progressive enrichment in O of the interstellar medium, BeB varies rather as the square of O, i.e. is of secondary nature. In the first case, fast nuclei enriched into He, C and O injected by supernovae and accelerated in surrounding superbubbles would explain the primary trend. In the second case, the main spallative agent would be the standard galactic cosmic rays. Calculated nucleosynthetic yields of massive stars, estimates of the energy cost of production of beryllium nuclei, and above all recent observations reported in this meeting seem to favor the primary mechanism, at least in the early Galaxy.