The astrophysical site of the r-process is still uncertain, and a full exploration of the systematics of this process in terms of its dependence on nuclear properties from stability to the neutron drip-line within realistic stellar environments has still to be undertaken. Sufficiently high neutron to seed ratios can only be obtained either in very neutron-rich low-entropy environments or moderately neutron-rich high-entropy environments, related to neutron star mergers (or jets of neutron star matter) and the high-entropy wind of core-collapse supernova explosions. As chemical evolution models seem to disfavor neutron star mergers, we focus here on high-entropy environments characterized by entropy $S$, electron abundance $Y_e$ and expansion velocity $V_{exp}$. We investigate the termination point of charged-particle reactions, and we define a maximum entropy $S_{final}$ for a given $V_{exp}$ and $Y_e$, beyond which the seed production of heavy elements fails due to the very small matter density. We then investigate whether an r-process subsequent to the charged-particle freeze-out can in principle be understood on the basis of the classical approach, which assumes a chemical equilibrium between neutron captures and photodisintegrations, possibly followed by a $\beta$-flow equilibrium. In particular, we illustrate how long such a chemical equilibrium approximation holds, how the freeze-out from such conditions affects the abundance pattern, and which role the late capture of neutrons originating from $\beta$-delayed neutron emission can play.