Tests

Except for the temporal arrangement of the data given which uses the field auf_dat from the table progression (verlauf), all information is gathered from the main table diabetik. Due to this, it was first tried to show that the temporal information taken from the table verlauf is not used correctly. According to the specification, only the last entry of each patient in each period of time had to be used. Hence, the following bugs seemed most likely:      

• A patient is being ignored in one period of time.
• A patient's data is inserted into a wrong period of time.
• If there are multiple entries in one period of time for one patient, the last entry is not used.
• If there are multiple entries in one period of time for one patient, an entry in addition to the last one is used.

It is not very likely that one of those faults has been made, because neither self made boundary value tests  nor a big volume test  produced any suspicious results.

Next, the produced information about the percentage of males and the percentage of foreigners were inspected. If a failure occurs during the determination of those two pieces of information, at least one of the following steps had to be wrong: counting the number of sex (geschl) fields which are equal to M, counting the number of  normal values apply (nw_gelten) fields  which are equal to n, and counting the total number of patients in the appropriate period of time. Again, the same methods which have been used to test the temporal arrangement of the data were unsuccessfully used to find bugs  in this part.

Thereafter the distinction of the three main types of diabetes mellitus  was inspected. This was only a little bit more complicated than the last two tests, because this time three different values instead of two were possible. It also had to be tested that the the different types of diabetes mellitus were classified correctly. Again, the program did not seem to do anything wrong.

The most interesting part of the patient's statistic was the  calculation of the number of days which lay between to given dates. This was necessary for the last two pieces of information calculated here. Especially, our calendar's leap-days  were a potential source for faults . A leap-day is inserted into our normal calendar every four years, or more precisely if the number of the year is divisible by four. Since this way too many leap-days are inserted, they are left out if the number of the year is divisible by 100, but not by 400. Equivalence class testing  was used for this part, but again no bugs could be detected.

Concluding the patient's statistics, some volume tests  were done which the module also passed.