I favour this approach too.
Andrew Schofield
On 01/10/2019, 09:15, "Ryanne Dolan" wrote:
Thanks Randall, that works for me.
Ryanne
On Tue, Oct 1, 2019 at 9:09 AM Randall Hauch wrote:
> Apologies for the late entry -- I entirely missed this KIP and discussion.
Thanks Randall, that works for me.
Ryanne
On Tue, Oct 1, 2019 at 9:09 AM Randall Hauch wrote:
> Apologies for the late entry -- I entirely missed this KIP and discussion.
> :-(
>
> Thanks for creating the KIP and proposing this change. I do think it's
> useful for source connector tasks to get
Apologies for the late entry -- I entirely missed this KIP and discussion.
:-(
Thanks for creating the KIP and proposing this change. I do think it's
useful for source connector tasks to get more information about the
acknowledgement after the record was written.
However, given the KIPs suggestio
Andrew, I have considered this, but I think passing null for RecordMetadata
would be surprising and error prone for anyone implementing SourceTask. I
figure the only use-case for overriding this variant (and not the existing
one) is to capture the RecordMetadata. If that's the case, every
implement
As you might expect, I like the overloaded commitRecord() but I think the
overloaded method should be called in exactly the same situations as the
previous method. When it does not reflect an ACK, the second parameter could be
null. The text of the KIP says that the overloaded method is only cal
