`offline' calibrations

For `offline' calibrations we know apriory the data validity periods. This is how they might look like Figure 3.

  

Figure 3:

In this example, calibrations cal₁,2 and 3 were done first, and then cal₄ was stored to superceed part of the cal₃ validity period. For the interval $]t_2 \ldots t_3[$ calibrations are absent.

Again, it is easy to see that one can make bad queries that ask for non-existant data or overlap multiple validity intervals. For example as in Figure 4

  

Figure 4:

A few side notes.

Notice that validity of cal₄ overlaps part of cal₃. If we could break cal₃ into two peices, to get rid of the overlap, but it would still be possible to make bad queries.

Notice also that we could extend this model by attaching labels to the calibrations. Let's say we call cal₁, cal₂ and cal₃ `default' calibrations and cal<sub>4</sub> `special Ian's calibrations'. In this extended model the getData() method would have an additional argument to specify `default' versus `special' calibrations. The bad query problem is still with us, even if we only query for `default' calibrations.