Hi Arvid,
> 1. `RelativeClock` looks like a super-interface of
> `org.apache.flink.util.clock.Clock`. Should we also reflect that
> accordingly by extending? This should not break anything.
It should be fine to do so, but to me the question is if we should do so?
It doesn't give any benefit
Hi Piotr,
thank you very much for addressing this issue. I'm convinced that the
approach is the right solution also in contrast to the alternatives.
Ultimately, only WatermarkGenratorWithIdleness needs to be adjusted with
this change.
My only concerns are regarding the actual code.
1.
Hi Piotr,
We've talked offline about this proposal and I think it would be beneficial
for users to get this fixed. +1 overall, and thanks for writing it down.
Best regards,
Martijn
On Tue, Jul 23, 2024 at 5:45 PM Piotr Nowojski wrote:
> Hi All,
>
> A bit unusual FLIP [1], as this is a bug
Hi All,
A bit unusual FLIP [1], as this is a bug fix for a problem that I have
recently discovered [2]. However I think FLIP is required, as properly
fixing the issue requires changes to the public API.
As this is a bug fix, I would propose to back-port this change to previous
releases (1.19 and