+1 (binding) -> Thanks for being receptive to all comments TP / Jason.

And regarding Jens' point: yes, "naming is difficult". However, at this
stage, this name is just a "codename" because it's "Java only,"
"experimental," mostly used internally (except for the package name in
configuration), and lacks a separate installable distribution (it's just a
Python package name). When/If we turn it (hopefully soon) into full-fledged
coordinators - with common APIs and a compatibility strategy—it **might**
get real "coordinator" features; this might get handy. It might also be
easier to "promote it" without migrations, which TP was rightfully
concerned about.

So, I actually like that it's named "coordinators" now in the Python
package name because it allows for easy future evolution without
unnecessary migration issues. I was far more sceptical about implementing
the new distribution naming schema at this point - because that would
"anchor" us much more. I think our discussion resulted in a good middle
ground: we avoid overcomplicating things (especially the development
process, operational complexity, and intra-compatibility issues), allowing
us to get something "working" quickly, while ensuring we aren't blocked and
have a smooth path to implement the longer-term vision later.

I think that was a very good discussion and outcome. Thanks again, TP.
Also, thanks to (a bit more silent in this discussion) Jason for being so
flexible. I really appreciate it. I know firsthand how difficult it is when
a bigger vision you have is kind of trimmed-down, and when you see where
you want to go and others seem to "not see it". It forces you to twist and
turn things to not lose the track of the bigger vision, while taking the
first baby step toward it. But my experience is that the end result might
eventually benefit from learnings along the way, so trimming the first
steps is a good thing (even if it's very difficult mentally). I've been
doing it for years in our dev environment. While it generally follows my
initial vision, it's very different now due to incremental steps and
tooling improvements along the way.

J.


On Sat, May 16, 2026 at 10:52 AM Shahar Epstein <[email protected]> wrote:

> +1 (binding), well done TP and Jason.
>
>
> Shahar
>
> On Sat, May 16, 2026 at 10:02 AM Tzu-ping Chung via dev <
> [email protected]> wrote:
>
>> Hi all,
>>
>> I’m calling vote on AIP-108: Java Task SDK and the Language Coordinator
>> Layer
>> AIP-108 Java Task SDK and the Language Coordinator Layer - Airflow -
>> Apache Software Foundation <https://cwiki.apache.org/confluence/x/pY4mGQ>
>> cwiki.apache.org <https://cwiki.apache.org/confluence/x/pY4mGQ>
>> [image: favicon.ico] <https://cwiki.apache.org/confluence/x/pY4mGQ>
>> <https://cwiki.apache.org/confluence/x/pY4mGQ>
>>
>> Discussion thread:
>> lists.apache.org
>> <https://lists.apache.org/thread/gjot4bxj9kygj2fk76kx6tyg8s4hr057>
>> [image: favicon.ico]
>> <https://lists.apache.org/thread/gjot4bxj9kygj2fk76kx6tyg8s4hr057>
>> <https://lists.apache.org/thread/gjot4bxj9kygj2fk76kx6tyg8s4hr057>
>>
>>
>> The vote will run for 5 days until Thursday, 21st May 2026, 07:00 UTC.
>>
>> Everyone is encouraged to vote, but only PMC members and Committers'
>> votes are considered binding.
>>
>> Please vote accordingly
>>
>> [ ] +1 Approve
>> [ ] +0 no opinion
>> [ ] -1 disapprove with the reason
>>
>> Consider this my +1 vote (binding)
>>
>> TP
>>
>

Reply via email to