Hi, there is no guarantee on the order in which the elements are processed. So it can happen that most elements from input one get processed before elements from the feedback get processed. In case of an infinite first input this will not happen, of of course.
For understanding what's going on it might also be helpful to print the parallel subtask index of the operator where you are printing output. For that you have to use a RichCoFlatMapFunction. In there, you can use getRuntimeContext().getSubtaskIndex() to know which parallel instance an operator is. Cheers, Aljoscha On Sat, 14 May 2016 at 19:55 Biplob Biswas <revolutioni...@gmail.com> wrote: > Can anyone help me understand how the out.collect() and the corresponding > broadcast) is working? > > > > -- > View this message in context: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Unexpected-behaviour-in-datastream-broadcast-tp6848p6925.html > Sent from the Apache Flink User Mailing List archive. mailing list archive > at Nabble.com. >
