From edb6426c79d49f237b2420ac070496a8594887d6 Mon Sep 17 00:00:00 2001 From: naman Date: Mon, 26 Oct 2026 01:13:39 +0530 Subject: [PATCH 1/2] Update the newsfragment eval cases to the policy from #74589 The cases have not changed since the harness landed in #69308, so they still encode the six-type policy. #74589 narrowed airflow-core to `significant` only -- "never for an ordinary bug fix, feature, improvement or doc change" -- which leaves the two cases that expect a feature newsfragment failing an agent for answering correctly. Flipping both would leave every case expecting "no", which an agent that always declines would pass, so two changes that do warrant a significant note keep the other direction covered: the asset-triggered run default from #68517 and the async Postgres driver default from #68496, both taken from newsfragments already on main. --- dev/skill-evals/cases/newsfragment.yaml | 32 ++++++++++++++++++++++--- 1 file changed, 29 insertions(+), 3 deletions(-) diff --git a/dev/skill-evals/cases/newsfragment.yaml b/dev/skill-evals/cases/newsfragment.yaml index b769afa54864a..0a9f14304cf62 100644 --- a/dev/skill-evals/cases/newsfragment.yaml +++ b/dev/skill-evals/cases/newsfragment.yaml @@ -63,7 +63,7 @@ - type: javascript value: 'output.should_create === false' -- description: "Worker connection testing (#62343): feature newsfragment" +- description: "Worker connection testing (#62343): no newsfragment" vars: request: | I added asynchronous connection testing: connection tests can now @@ -74,9 +74,9 @@ Should I create a newsfragment? assert: - type: javascript - value: 'output.should_create === true' + value: 'output.should_create === false' -- description: "Coordinator layer (#65958): feature newsfragment" +- description: "Coordinator layer (#65958): no newsfragment" vars: request: | I added a coordinator layer to the task execution supervisor so @@ -84,6 +84,32 @@ along with a first Java coordinator and a new config option. Most of the changes are in task-sdk, with some in airflow-core. Should I create a newsfragment? + assert: + - type: javascript + value: 'output.should_create === false' + +- description: "Asset-triggered Dag run default (#68517): significant newsfragment" + vars: + request: | + I changed single-asset schedules and asset OR expressions to create + one Dag run per event instead of consuming all queued events in one + run, which also changes AssetOrTimeSchedule and + PartitionedAssetTimetable. Existing Dags scheduled this way will + start producing a different number of runs after upgrading. + Should I create a newsfragment? + assert: + - type: javascript + value: 'output.should_create === true' + +- description: "Async Postgres driver default (#68496): significant newsfragment" + vars: + request: | + I changed the driver Airflow derives for the async metadata database + when [database] sql_alchemy_conn_async is unset: for PostgreSQL the + derived URL now uses psycopg3 instead of asyncpg when psycopg3 is + installed. Deployments that did not set the async URL explicitly will + connect through a different driver after upgrading. + Should I create a newsfragment? assert: - type: javascript value: 'output.should_create === true' From ba7f6d8772a87cda55548236903093987e2c3c14 Mon Sep 17 00:00:00 2001 From: naman Date: Wed, 28 Oct 2026 18:21:14 +0530 Subject: [PATCH 2/2] Refresh the skill-eval hash for the updated newsfragment cases --- dev/skill-evals/last-eval-hash.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/dev/skill-evals/last-eval-hash.txt b/dev/skill-evals/last-eval-hash.txt index f489ded32f581..fe4ae9b9d6781 100644 --- a/dev/skill-evals/last-eval-hash.txt +++ b/dev/skill-evals/last-eval-hash.txt @@ -1,3 +1,3 @@ # Generated by dev/skill-evals/eval.py — do not edit or resolve conflicts by hand. # Run `prek run run-skill-eval --hook-stage manual --all-files` to regenerate. -32e8c893747492d95d98e47b9244b693f9b23756c9d7502cd6aae94225e4ef27 +8bbe0b35079bdb9d66a87ba88badf083bc8ffb3403fc89468d8ae099fec5686b