For many reasons, either because of the measurement methods or data
recording techniques, data might be only available as discrete data point
objects.
For example in MS spectra, data points are only available for detected
masses. Data points in such MS spectra are not equidistant at all and
it would not make sense to have them in an equidistant data object. Some
of the modern recording techniques, like diode array detectors, provide
data in a nearly equidistant data sets, but there are very small deviations
in the data point distance because of the detector precision, which lets
such data occur as discrete data objects as well.
In detail this means, there is no continuous spectrum with regular data
point intervals at least on one axis, but only a list of data points with
random distance belonging together.