Hmm Thanks Doesn’t the count = 2 cause a problem though ? I mean one stream may be considerably faster than the other (or am I getting something wrong)
The aggregation you mentioned is actually the same as what I thought (i.e. with window) Couldn’t find in the docs how you would do aggregation without window
i think it's more the high-level flow that roland is talking about -- how you model the underlying aggregate/window is up to you, and can accommodate for these things
i think @roland.thiolliere might be right that a session window makes sense, but like i said, i haven't worked with windows a lot
Thanks again