Nice. We should have more of those. "Yeah, good idea to drop it" rather
than AIPs lingering about"

The recent restructure of AIP pages makes ite easier to see where we are -
but an AIP that author knows we will not likely work on should really close
it.

Thanks Shahar!

On Fri, Nov 13, 2026 at 5:31 PM Shahar Epstein via dev <
[email protected]> wrote:

> Hi all,
>
> I am withdrawing AIP-106.
>
> Following the discussion in this thread and comments on the AIP, I explored
> success thresholds on mapped tasks, as Jarek suggested. I've came into the
> conclusion that the benefits do not justify the added complexity.
>
> Ash questioned why the downstream task could not use all_done and check the
> results itself, and whether upstream_failed rather than failed justified
> scheduler changes.
>
> I tested an all_done gate task locally - it checked the success count and
> passed surviving results to both ordinary and mapped consumers. This
> covered the use cases I was trying to address without scheduler changes.
>
> I moved the AIP to the "Abandoned" section on the wiki - if anyone wants to
> revisit it in the future - you're more than welcome.
>
> Thank you to Jarek, Ash, Amogh, Vikram, and Jens for the feedback.
>
>
> Shahar
>
> On Tue, Apr 21, 2026 at 2:00 PM Shahar Epstein <[email protected]> wrote:
>
> > Hey everyone,
> >
> > I'd like to get your feedback on a new AIP proposal:
> >
> > AIP-106: Composable Trigger Rules
> > https://cwiki.apache.org/confluence/x/64wmGQ
> > <
> https://cwiki.apache.org/confluence/display/AIRFLOW/AIP-106+Composable+Trigger+Rules
> >
> >
> > tl;dr
> >
> > This proposal introduces a composable expression model for trigger rules,
> > enabling conditions not expressible with the current enum-based system,
> > while fully preserving existing trigger rules for simple cases.
> > For example, "All done, at least two succeeded" can be expressed as
> > TR.expr(done="all", success=">=2").
> >
> > Design Notes/Constraints:
> > 1. The aggregate form is AND-only by design, to preserve determinism and
> > keep evaluation predictable.
> > 2. The composable model complements (rather than replaces) the enum. Some
> > trigger rules (e.g. ALL_DONE_SETUP_SUCCESS, ONE_DONE, ALWAYS) remain
> > intentionally out of scope, and enums remain the more readable option for
> > simple cases.
> >
> > I'd appreciate your feedback on:
> >
> >    -
> >
> >    Backward compatibility with existing trigger rules (see Appendix A)
> >    -
> >
> >    Semantics of eager vs complete evaluation
> >    -
> >
> >    Scope and limits of the expression model
> >    -
> >
> >    Potential edge cases and performance implications
> >    -
> >
> >    UI/observability considerations for debugging expressions
> >    - Anything else
> >
> >
> > Thank you!
> >
> >
> > Shahar
> >
>

Reply via email to