I'm unable to find the best way (the drools fusion way actually) to count "A" events between "B" events (STREAM mode). An example of sequence could be:
B A A A B A A B A...
A use case could be
verify a sequence validity i.e there is always 3 "A" events between each "B" event
I don't know if the "window over" concept is appropriate in such a situation. Thanks for your help, I still have some difficulties to think in CEP way.
Two Drools versions cause two different behaviors (same Unit Test). So I think my misunderstood comes from a bug in "windows" behavior. I can't be more specific this time.