Jan Lukavský
Jan Lukavský
I'm not sure if this is directly related to this - but there is at least one valid use case of stateful non-windowed operation - that is a non-windowed stream...
I agree there is a strong connection between my remark and the issue #41. The main difference that I see here, is that it is effectively possible (under some circumstances)...
Thanks for the report. Does the provided test fail?
If so, can you please create branch with this failing test?
There is fundamental flaw in translation of Flow into runnable DAG. This flaw will be solved by #256, until then I suggest (a slightly suboptimal and highly ugly) workaround: -...
Is this still relevant?
Just a suggestion, it would seem reasonable to divide the performance comparison into two subtypes of jobs - CPU bound and IO bound.
This looks good! :+1: Is the first figure the state before the start of the optimization? How is spark executor performing after the optimizations compared to the raw Spark? Is...
My bad, I somehow ignored the descriptions of the figures. :) The overhead of flink batch executor comes from the (de)serialization during sorting issue? Anyway, these improvements are really promising....
Good to hear! :+1: