Hi Damain,
The semantics of Count() is still depending on the KTable, where the
records are of changelogs based on key. So for your example the KTable is
really conceptually from an updating "source table" with only one record
but with updating values from 1 to 5. And KTable.count(...) is essentia
Hi Liquan,
Thanks for getting back to me and pointing me to the confluent doco. Based
on what i read and my own assumptions, i'd expect the data consumed from
the output topic to be:
A:1
A:2
A:3
A:4
A:5
What am i missing?
Thanks,
Damian
On Sun, 17 Apr 2016 at 19:20 Liquan Pei wrote:
> Hi Dam
Hi Damin,
I am new to KStreams as well, so my answer might not be 100% precise. In
KTable, the same key is treated as updates instead of events. Thus
aggregation on the same key will do some de-dup. The docs for the tech
preview contains some explanation on this behavior:
http://docs.confluent.io
