next up previous contents index
Next: Acquisition of Achievements Up: Testing DPV Previous: Testing DPV

About all Tests

  It is not possible to start testing right away. It is necessary to do some preparations. The first thing done was to collect the important information given in the manual. Then a possible algorithm  to calculate the statistic  was determined. This includes to ascertain the fields of the database  needed. This information was checked by alternating both the appropriate fields and those that are unlikely to be used.

Next the module was reprogrammed in   perlgif. After testing the new programs with inputs that are small or easy to evaluate by hand, the ``real'' tests were started. If a fault was suspected during a test, the input was  simplified until it was possible to decide if it was in the dBASE  or the  perl module.

The first test done was always the inspection of the given data's temporal arrangement. In other words, is each medication taken into account in the appropriate quarter, and in the appropriate year, and nowhere else? To do this, some small dBASE  tables were produced with just a few entries. After those were evaluated correctly, the table's size was continuously increased. The correct response to empty periods of time was also checked.



Ingo Melzer
Mon Aug 5 15:12:01 MET DST 1996