depending on the implementation in `PipelinedSubpartition`, i.e. `if 
(buffers.size() == 1 && buffers.peekLast().isFinished())` or whatever we change 
it to (we don't make guarantees here!), the producer thread may not have 
flushed its last record after finishing and the source would wait forever (no 
output flusher in that test)
-> we need to flush all channels before leaving the producer

[ Full content available at: https://github.com/apache/flink/pull/6692 ]
This message was relayed via gitbox.apache.org for [email protected]

Reply via email to