1fanwang opened a new pull request, #66324: URL: https://github.com/apache/airflow/pull/66324
related: #66323 `failed` and `upstream_failed` currently render with ΔE2000=6.2 in normal vision — below the "easily distinguishable" threshold of ~10. In a Grid view they look like the same color, especially in dense state mixes. Issue #66323 has the full simulation data; this PR is the smallest fix for the immediate normal-vision regression. `upstream_failed` switches from `orange.600` (hue ~36°) to `amber.600` (hue ~58°), pushing it clear of `red.600` (hue ~27°). The `amber` palette is already defined in `theme.ts` so no new tokens are introduced. ### Before / after (Brettel/Machado simulation, ΔE2000) | pair | before normal | after normal | before deutan | after deutan | |---|---:|---:|---:|---:| | failed vs upstream_failed | **6.2** | 20.9 | **0.4** | 8.3 | | upstream_failed vs up_for_retry | 24.8 | 10.8 | 9.1 | 1.2 | Normal-vision `upstream_failed` vs `up_for_retry` drops from 24.8 to 10.8 — still above the 10 threshold, so no new normal-vision regression. The deuteranopia collapse of `upstream_failed` vs `up_for_retry` (8.3 → 1.2) is part of the broader colorblind story in #66323. The warm-hue band structurally collapses under deuteranopia regardless of which OKLCH is chosen — fixing that fully needs either a per-user CB-safe preset (Okabe-Ito-derived, what GitHub Primer ships as four theme variants) or a non-warm hue for one of these states. Both are bigger conversations. Filing as a draft to invite discussion. Reasonable alternatives: - this PR's choice — `amber.600`, simplest swap. - tune the `orange.600` OKLCH directly to a higher hue. - bundle with broader CB work — ship Okabe-Ito preset + extend `ThemeColors` (#64232 pattern) to expose state-color tokens. Happy to take it in any direction. If `amber.600` is fine, will mark ready, add a newsfragment + tests. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
