Hi Mathieu,
You are correct in that the de-duping only occurs within the commit
interval.
I can understand and appreciate the use-case you have. So I think the right
approach for this is to create a KIP with your suggested changes and put it
to the community. Are you happy to do that?
Thanks,
Da
Hi Eno,
I'm not sure. My understanding is that the cache would prevent two
immediate updates for the same key from being forwarded, but that only
applies when records arrive within commit.interval.ms of each other. Is
that understanding correct?
filterRedundant compares the newValue & oldValue
Hi Mathieu,
Thanks for the suggestion. Wouldn't the cache introduced in KIP-63 do some of
this for you, in that it dedups records with the same key and prevents them
from being forwarded downstream?
Eno
> On 4 Dec 2016, at 04:13, Mathieu Fenniak wrote:
>
> Hey all,
>
> I'd like to contribute
