`online' calibrations

Let's consider the `online' calibrations first. Figure 1 shows three different consecutive calibrations ( cal₁, cal₂, cal₃) stored at time ( $t_1, t_2, t_3, \ldots$). The validity intervals are well defined. Between t₁ and t₂ cal₁ is valid, between t₂ and t₃ cal₂ is valid and from t₃ to infinity, cal₃is valid, until superceeded by cal₄ stored at a later time t₄.

  

Figure 1: `Online' calibrations

Let's take a look at calibrations retrieval.

We have already decided that the getData() function will specify a time interval $[t_{start}\ldots t_{end}]$ and the database will return the calibrations that are valid inside this interval and that the application can safely use them without further validity checks or getData() calls. t_start and t_end might correspond to the run start and end times or to the times of the first and last events in a data file.

Examples are in order (see Figure 2).

  

Figure 2: Querieng `online' calibrations

query₁ and query₂ return the obvious results.

query₃ is more interesting because it happens to overlap two validity intervals. What should such a query return? It cannot return cal₁ because it is not valid in the interval [t₂..t_end], it cannot return cal₂ because it is not valid in the interval [t_start..t₂]. Therefore it should return an error because it cannot return any data. I would call this case a bad query because obviously there is nothing wrong with the data in the database and yet the query cannot return any data.

This example shows that even the simplest validity interval scheme has to deal with bad queries.