The result file formats support two kinds of results: output vectors and
output scalars. Output vectors are basically timestamped numeric data, used
to record the history of one particular simulation run. Output scalars are
summary data, usually written out at the end of the simulation. Scalars can
be simply numeric, or can be some statistical summary (mean, std.dev, min,
max, histogram data, etc).

It is assumed that the simulation model is composed of components (or in
OMNeT++ terms, modules), and that every simulation result (scalar or
vector) is generated by, i.e. belongs to, some component. (Global
statistics can be thought of belonging to the "root" component.) A
simulation result is identified by the name of the component
("aloha.host2") plus a result name ("numPacketsSent"). Component names will
be usually (but not necessarily) hierarchical; the recommended separator
between component path segments is ".".

Every simulation result can be annotated with metadata. Metadata
annotations are called attributes; two examples for attribute usage are
"unit" (physical unit, e.g. "mW") of some result, and "interpolation-mode"
(linear, sample-hold, etc., for vectors) to serve as a hint for plotting.

Simulation results are generated during simulation runs. Info about
simulation runs is also recorded as a header into the result files. Every
simulation run is identified by a runID, which is intended to be
"reasonably globally unique" identifier (concatenated from a number of
fields like hostname and datetime, to be practically unique but still
meaningful and not too long). Further run attributes are defined to record
data like the name of the model, its parameters, the random number seeds,
various simulator configuration settings and so on. The intent is to record
enough information so that the user can later reproduce the results.

Three special-purpose run attributes, "experiment", "measurement" and
"replication", are related to organizing record parameter studies.