Hi Piotr,

The reasoning makes sense to me. I probably meant to say `public` class
visibility and `@Internal` in my proposal, similar to other
connector interface components (e.g. KafkaSourceReader).

However, I actually found a workaround using the Public API sink builder
and factory classes to create the sink writer, so I will abandon this
effort for now until it would be required. Thanks for the feedback!

Best,
Mason

On Tue, Aug 30, 2022 at 7:00 AM Piotr Nowojski <pnowoj...@apache.org> wrote:

> Hi Mason,
>
> I would be against making them part of the Public API. It would be against
> our current conventions and it would make future changes in this connector
> much more difficult. It's difficult to support Public APIs, even if they
> are `@PublicEvolving`.
>
> Best,
> Piotrek
>
> pon., 29 sie 2022 o 19:17 Mason Chen <mas.chen6...@gmail.com> napisaƂ(a):
>
> > Hi all,
> >
> > I know everyone is busy with preparing the 1.16 release. Just want to
> give
> > this thread a gentle ping, please give your +1 if there are no concerns!
> >
> > Best,
> > Mason
> >
> > On Thu, Aug 11, 2022 at 1:27 AM Mason Chen <mas.chen6...@gmail.com>
> wrote:
> >
> > > I understand that the 1.16 feature freeze is in place so I propose for
> > > these changes to go in the 1.17 release.
> > >
> > > Best,
> > > Mason
> > >
> > > On Thu, Aug 11, 2022 at 1:20 AM Mason Chen <mas.chen6...@gmail.com>
> > wrote:
> > >
> > >> Hi devs and community,
> > >>
> > >> I would like to discuss making the low level components of the
> KafkaSink
> > >> PublicEvolving. It looks like all of the classes are package private
> > apart
> > >> from the public facing API and classes were put in the same package to
> > make
> > >> this work. My desire to make at least the KafkaSinkWriter a public
> > class,
> > >> with no further changes to private/package-private methods. The
> > class(es)
> > >> would be annotated with PublicEvolving.
> > >>
> > >> I am in the process of implementing my own Sink that writes to Kafka
> and
> > >> I want to implement the SinkWriter by composing the functionality
> > contained
> > >> in the KafkaSinkWriter. There is significant work in this
> implementation
> > >> and I would like to use it and contribute back, bug fixes for example,
> > if I
> > >> find any in independent testing.
> > >>
> > >> Hopefully, that motivates the request. Please let me know your
> thoughts!
> > >>
> > >> Best,
> > >> Mason
> > >>
> > >
> >
>

Reply via email to