Plot period is broken, or means something different than KaSim's

[hmedina]

The central idea behind trace replaying is the capacity to compare stuff across multiple traces.

• Option -p plot period has a confusing behavior. E.g. -p 1 does not yield an observation point (roughly) every time unit, but closer to an observation per event. Moreover -p 1000 seems to approach a period of 1 observation per time unit, at least late in the simulation. Why 1000? Is the period relative to the total length of the trace?

• If the user wants a plot point every time unit, the user should get that, not roughly every time unit. The key difference is that 2 traces will almost never fall on the same float value of time. These should be dumped in alarm time, or at least like KaSim does respecting the plot period.

Usecase: I have a model with a parameter that can be any of m values. I want to compare the giant-component generation by these m values. I simulate each parametrization for n time units with plot period of 1, get m output files each with n values nicely indexed by their n integer time points. Then I switch to ReConKa and now all my connectivity.csv values across parametrizations are not comparable (i.e. the first point is different for every parametrization). With -p 1 I would expect ReConKa to produce a file with n rows, like KaSim does.