Evaluating KPI by Process Interpreter

• KPIs are evaluated by Process Interpreter at the end of each step, i.e. after all data objects are already populated by script tasks and by output data mappings from activities. Only KPIs whose trigger data object value has changed by the output data mapping, which is indicated by data association in a process diagram, are evaluated.
• If there is an exception during KPI python expression evaluation, the stack trace is stored in "exlog" attribute of KPI value and KPI value is set to null. Error during KPI evaluation does not affect state of a process instance.
• In a contrary to data object, which has only the current value (previous values are forgotten), each result of KPI value evaluation is archived. It allows e.g. to record KPI values of KPI defined within a loop. The values of the same KPI are differentiated by the stepNo attribute of KpiValue. When KPI is evaluated the first time, stepNo is set to 1, subsequent evaluation of the KPI value sets stepNo = stepNo + 1.
• When a trigger data object is not set in the script, not even by output data mapping, then the corresponding KPI is not evaluated.