In knowledge representation, currently, there exist two main approaches to representing temporal relations between events: the approach that uses moments of time, and the approach that uses time intervals. In real life, we encounter both instantaneous events (that correspond to moments of time) and non-instantaneous ones (that correspond to intervals). Therefore, the ideal knowledge representation language should describe both moments and intervals. It is, of course, possible to describe moments of time in interval-based formalisms as degenerate intervals and, correspondingly, intervals as pairs of moments. However, this artificial representation leads to unnecessary computations: e.g., when we represent each moment of time $t$ as a degenerate interval $[t,t]$, then every operation with moments of time requires two (identical) operations with interval bounds.
In this paper, a new formalism is described that combines points and intervals and that has a good computational behavior.