The electronic stopping power is a crucial quantity for ion irradiation: it governs the deposited heat, the damage profile, and the implantation depth. Whereas experimental data are readily available for elemental solids, the data are much more scarce for compounds. Here we develop a fully ab initio computational scheme based on linear response time-dependent density-functional theory to predict the random electronic stopping power (RESP) of materials without any empirical fitting. We show that the calculated RESP compares well with experimental data, when at full convergence, with the inclusion of the core states and of the exchange correlation. We evaluate the unexpectedly limited magnitude of the nonlinear terms in the RESP by comparing with other approaches based on the time propagation of time-dependent density-functional theory. Finally, we check the validity of a few empirical rules of thumbs that are commonly used to estimate the electronic stopping power.