Wolfgang,
I think I have found a fix for the issue you reported. Anyway, it is not
very clear to me what you originally wanted to achieve with this example
since, by using my fix, the 2 rules cause an endless loop even if you used
the no-loop flag because each one feeds the other. I think this is the
correct behavior unless you don't annotate the MySet class with
@PropertyReactive. Indeed I checked that making the class property specific
allows you to avoid the infinite loop and the test terminates with the
following 2 sets:
[z, y, x]
[two, one, three, four, z]
Please confirm that what I am reporting corresponds with your expectations.
I also noticed that, in the test you attached, you used the mvel dialect
for the first rule and by doing that the drl compilation fails. At a first
glance this is caused by a mvel bug, not supporting varargs in the
constructor signature. I am going to give a look also to this problem.
Mario