Files
windmill/frontend/src/lib/components
Ruben Fiszel 6ae8dd37b1 fix(pipelines): pipeline-level run control, tables label, data-test rollback + fork badges (#9944)
* fix(pipelines): pipeline-level run control, tables label, data-test rollback + fork badges

- Add always-visible "Run pipeline" header control (edit mode) that runs
  every script in dependency order via the bounded-cascade engine, so a run
  no longer requires hovering a node's play button.
- Header summary counts ducklake/datatable assets as "tables" (and s3object
  as "files") instead of the raw kind, collapsing shared nouns.
- Surface a data-test outcome badge on guarded asset nodes: EE shows a
  rolled-back (previous version left live) state, CE shows published-despite-
  failure — driven by the producer's last run state and the edition.
- Make the fork data-environment marker a prominent labeled chip
  (⑂ fork / ↗ parent) instead of a bare icon.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(pipelines): address CI review — scope Run pipeline to members, anchor guard badge, spin loader

- Run pipeline now filters to `in_pipeline` script runnables, so it never
  launches dependency-only endpoints the graph shows for context (macro
  libraries, custom data-test scripts, out-of-folder producers).
- Data-test guard badge only attaches to the producer's declared
  `// materialize` target, so a multi-output producer no longer badges its
  other ducklake writes.
- Spin the Loader2 icon in the "Run pipeline" button while a run is in
  progress (startIcon classes), matching every other loading affordance.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(pipelines): data-test badge copy speaks to write policy, not failure cause

producerFailed is a generic job-failure signal, so the failed-state tooltip
no longer claims the run "failed its data tests" (it could be a runtime/worker
error). It now states the edition's behavior on any failed materialize: EE
rolls back (previous version left live), CE may leave a failing write live.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(pipelines): Run pipeline keeps independent branches running after a failure

runSelection used a single global fail-fast flag, so once any node failed it
refused to schedule *any* newly-ready node — a failure in one branch could
strand an unrelated healthy branch as 'skipped' depending on job timing. Now a
failure poisons only its transitive descendants; independent branches finish.

Add regression tests: independent-branch-survives-failure and join-node-skipped
-when-one-upstream-fails.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-06 00:08:24 +02:00
..
2025-09-11 07:57:01 +00:00
2026-06-22 09:33:55 +02:00
2026-03-05 18:11:40 +01:00
2025-10-30 06:47:26 +00:00
2025-10-28 17:52:04 +00:00
2026-03-05 18:11:40 +01:00
2025-11-04 19:13:09 +00:00
2025-11-29 01:23:47 +00:00
2026-03-05 18:11:40 +01:00
2025-11-04 19:13:09 +00:00
2026-02-11 06:27:21 +00:00
2026-02-19 09:33:33 +00:00
2025-11-29 01:23:47 +00:00
2026-03-05 18:11:40 +01:00
2026-02-19 09:33:33 +00:00
2026-03-05 18:11:40 +01:00
2026-03-05 18:11:40 +01:00