Hi all,

When we introduced airflowctl under AIP-81 / AIP-94 [1][2], the stated
direction was to reach 1.0.0 "step by step". We shipped 0.1.0 in Nov 2025
and are now at 0.1.5, with broad feature coverage and a CLI grammar that
has settled (required params positional, optional ones as flags, per the
parameter-style lazy consensus [3]). I think it is time to take that step
and release 1.0.0.

Concretely, 1.0.0 would put the user-facing CLI surface under SemVer, with
deprecations following the same policy as Airflow itself.

One more thing worth discussing is the coupling with core. Airflow core now
depends on apache-airflow-ctl with a pinned version range, and the client
datamodels are generated from the core API. Jarek suggested earlier that
releasing the two together could make this easier with TaskSDK, which feels
worth considering as it can provide strict versioning between releases but
easier maintenance between release cycles to one catch another all the
time. Therefore, I would like to hear what people think about keeping
airflowctl on its own cadence versus aligning the 1.0.0 cut with a core
release. This could also help combine release manager effort and reduce
overall release coordination overhead.

We plan to start the release vote at the end of June, so please share any
objections or thoughts before then. The additional combined distribution
part is not a hard dependency. We can proceed with a couple of iterations
and evaluate the results as we go. If there is no strong agreement on this
idea, we will continue with separate release cycles.


[1] AIP-81:
https://cwiki.apache.org/confluence/display/AIRFLOW/AIP-81+Enhanced+Security+in+CLI+via+Integration+of+API
[2] AIP-94:
https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=382175838
[3] Parameter style lazy consensus:
https://lists.apache.org/thread/m1qvcvow3l17ytv40vhslh40wn3rntrm

Thanks,
Best regards,
Bugra Ozturk

Reply via email to