The 'background edit from empty state' test matched the rundown row by the
shared prefix '0 empty-'. Each Playwright retry creates another empty-<ts>
rundown, so on retry the locator resolved to multiple rows and failed with a
strict-mode violation — preventing the (pre-existing, timing-related) flake
from ever self-healing. Scope the locator to the unique name the test just
created so each attempt is independent and retries can recover.
The lazy editor (Exp A) unmounts on exit, so the editor's own deferred
focus-return (useReactiveTextInput) targeted a detached node and focus was
lost to <body>, breaking table keyboard navigation after submit/escape.
Return focus to the parent cell deterministically in a layout effect once the
display element is back in the DOM (rather than a setTimeout that races the
next keypress), mirroring the time/duration cells.
Update the e2e tests to the lazy interaction model (click the cell to open the
on-demand editor, assert it unmounts on exit) — matching the existing timer
cell test. Verified end-to-end: 202-cuesheet (3/3) and 206-url-preset (5/5).
* feat: create the first event from the cuesheet
* chore: cleanup css
* test: cuesheet beckground edit from empty state
* feat: only show +event button when user is in edit mode and have full write perms
The metrics scaffold has served its purpose: it proved the lazy text-cell
editor change (Exp A) and ruled out the colour-precompute and lazy-autosize
candidates. Removing it entirely:
- restores AutoTextarea/Tooltip/useReactiveTextInput/EventRow/CuesheetTable
to their original (uninstrumented) form
- deletes common/devtools/cuesheet-metrics
Keeps the validated Exp A optimization (lazy editor mounting). Verified:
typecheck, lint, 181 unit tests, and production build all pass; no scaffold
references remain in the bundle.
Render a lightweight, focusable TextLikeInput for cue/title/note/custom cells
and only mount the real input/textarea editor (with reactive-input hooks and
autosize) on click/focus, mirroring the existing time/duration cells.
Measured baseline vs Exp A at CPU 4x (median of 3, ~1140 row mounts):
- longFrames 201 -> 127 (-37%), maxFrameMs 517 -> 417 (-19%)
- longtask total 58.9s -> 30.7s (-48%), EventRow.mount avg 179 -> 146ms (-18%)
- cell.autosize and cell.reactiveInputInit eliminated from the scroll path
- per-row re-renders 4648 -> 3048 (-34%): stable display height, no virtuoso churn
(the opposite of the reverted lazy-autosize Exp B, which doubled re-renders)
Editing verified end-to-end: click mounts+focuses the editor, edits persist,
editor unmounts back to the display on commit. TextLikeInput now fills its cell
so the whole cell is clickable to edit.
- Scaffold: snapshot()/getResults() so the benchmark returns structured
metrics for automated before/after capture (PERF-METRICS).
- Candidate Exp B: defer AutoTextarea autosize() off the mount path to
focus-time, removing the per-row forced reflows during scroll. Under
evaluation against the metrics.
Adds a dev-only, self-contained instrumentation module under
common/devtools/cuesheet-metrics to measure the first-render (mount) cost
of virtualised cuesheet rows, so scroll-performance changes can be backed
by before/after numbers.
- Console-only API on window.__cuesheetPerf (deterministic scroll benchmark
+ manual start/stop), gated by isDev && ?perf=1, runtime-inert in prod.
- Whole-row mount probe plus per-subsystem attribution: colour calc,
getVisibleCells, AutoTextarea autosize, Tooltip portal mount,
useReactiveTextInput hotkey-handler init; FPS/long-frame timing.
- All edits to existing files tagged // PERF-METRICS for a one-pass teardown.
Temporary: to be deleted once the optimizations are proven.
* refactor: allow secondary rundowns
* ui: move dropdown
* feat: follow loaded
* ui: background edit warning
ui: fix disable radio button
* feat(ui): add loaded sufix in the rundown list
* feat(ui): add direct link to background edit from rundown manager
* fix: better fallback
* fix: default to is isCurrentRundown for nav bar colour
* chore: rename navigate to cuesheet
---------
Co-authored-by: alex-arc <ac@omnivox.dk>
Adds timer-legacy.html — a zero-dependency static page that connects
to the Ontime WebSocket and displays the live countdown for browsers
that cannot parse the Vite-built React bundle (Safari < 14, Chrome < 85).
Also adds a synchronous pre-React guard in index.html that redirects
/timer visitors to the legacy page when Promise.any is unavailable,
matching the Vite build target floor of Safari 14 / Chrome 85.
https://claude.ai/code/session_013MfdwdjdDUnr3akpGSGXa4
* feat: update auto cue numbering
* feat: renumber from ui
* refactor: patchEntries is not used
* chore: format
* fix: correct cue at top of group
* fix: handle precision
* refactor dialog
* bump limit for performance time test
* extract type
* add class name to lable
* fix rebase
* refator: extract renumering logic
* chore: comments for getIntegerAndFraction function
* chore: add the for renumber mutation
* fix: fraction match precision
* refactor: small cleanup
* refactor: use more narrow validator
---------
Co-authored-by: alex-Arc <omnivox@LAPTOP-RC5SNBVV.localdomain>
* refactor: remove unneeded async
* chore: add express Router type to all routes
* refactor: don't use index in react key
* refactor: correctly get error message in excel route
* refactor: avoid exporting muteable values
* refactor: replace traslation hook refetch time with stale time
* refactor: make it the service responsebillety to send refetch keys for asset changes
Claude
Alex Christoffer Rasmussen
Carlos Valente
Carlos Valente
Cameron Slipp
Joel Wetzell
Joel Wetzell
Brian Rodgers