utils/unified-design-system/nielsen-audit.md
Krzysztof kuhy Rudnicki cae0870fd0 Add the Nielsen coverage audit alongside design_system v0.1.0
The grid maps Nielsen's 10 GUI elements to what actually exists per stack,
and records the deferred clusters (GitHubMirrorScreen x5, SettingsScreen x6,
~20 list tiles, filter sheets) with repo, path and line count so none of it
needs rediscovering.

It also records three corrections the measurement turned up, because each
one changes what is worth doing next:

- The "343 raw color literals" are not spread through call sites; they are
  concentrated in the hand-copied theme files. todo had 35, all 35 inside
  theme.dart. Seven repos have zero literals outside their token file, so
  deleting it removes essentially all of them. The repos with genuine
  call-site leakage are billsplit, epopeja_karta and focus_owner.
- There are 4 AppStatusColors, not 5, and they disagree about their own
  field sets; the package's version is a superset so nobody loses a field.
- There are 3 section headers, not 5, and no _SectionLabel exists.
  epopeja_karta's is a different element (an eyebrow, not a section title)
  and was deliberately left out rather than force-fitted.

Also records that workout_app, cited as a consumer in the plan and in
several skill docs, does not exist on disk.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XS1m7b1piBtAL22USLW6jL
2026-08-14 23:01:31 +02:00

11 KiB
Raw Blame History

Nielsen coverage audit

Jakob Nielsen's "10 GUI Design Elements Build Every User Interface" argues that essentially every UI is assembled from ~10 standardized elements, and that the value comes from reusing those conventions rather than re-inventing them. Re-implementing the same button in eight repos is the private version of the convention erosion he describes.

This file is the audit grid, not a build list. Shipping a widget nobody imports is the failure mode to avoid — 183 imported agent definitions were invoked zero times in 432 sessions. A component gets built only where two or more repos already contain a structurally similar implementation.

Written 2026-08-14, alongside design_system v0.1.0.

Scope

Fourteen Flutter repos, seven web repos, four Python/Tk guard apps. Out of scope, verified rather than assumed: Unity (zero .uxml/.uss files exist anywhere), games (their repeated identifiers are game constants like FORGE_MAX_C, not UI), and the locker gate windows (their X-grab/input-hijack invariants make the window non-reusable, though its leaf widgets are fair game).

~/testsAndMisc/pomodoro_app and horatio no longer exist. workout_app does not exist either — it is named in the source plan and in several skill docs, but no directory under ~ matches *workout*. Anything citing it as a consumer is stale.

The grid

# Nielsen element Flutter Web Python/Tk
1 Buttons ⚠️ themed only ❌ 7 vocabularies ⚠️ 4 factories, see Phase 2
2 Input fields / forms ✅ inputDecorationTheme ❌ ⚠️
3 Menus ❌ not specced ❌ ❌
4 Links ❌ not specced ❌ n/a
5 Dialogs ✅ confirmDestructive() ❌ ⚠️
6 Alerts / notifications ✅ showToast/showError ❌ ❌
7 Icons ❌ not specced ❌ ❌
8 Checkboxes / radio buttons ⚠️ themed only ❌ ❌
9 Tabs ❌ not specced ❌ ❌
10 Search ❌ not specced ⚠️ FilterBar ×2 n/a

components.html specs 9 components but says nothing about menus, links, icons, radio buttons, tabs, or search — six of Nielsen's ten. That is the documentation gap; whether any of them deserves code still depends on the two-repo extraction criterion, which most of them currently fail.

What v0.1.0 closed

Export Replaced
AppPalette + scales 10 divergent theme.dart copies; ~318 raw color literals
buildLightTheme / buildDarkTheme the same, per repo
AppStatusColors 4 hand-rolled ThemeExtensions with inconsistent fields
confirmDestructive() ~35 inlined AlertDialogs
showToast / showError ~38 raw SnackBar sites, zero helpers
EmptyState 4 variants, one reusable
SectionHeader 3 near-identical private widgets

Proof consumers todo and untools are migrated, verified and pushed.

Corrections to the original plan, found by measuring

The plan's figures were mostly right but three were not, and the difference changes what is worth doing next:

  • The "343 raw literals" are not spread through call sites. They are concentrated in the hand-copied theme files. todo had 35, all 35 inside theme.dart; untools had 36, of which 35 were in theme.dart and the last was Colors.transparent in keyboard.dart — a sentinel, not a palette value. Deleting the theme file removes essentially all of them. There was no call-site sweep to do in either proof consumer, and the same is true of home_inventory, habit_stack, diet-guard, wake-alarm, macro-cam and kuhylog, which all have zero literals outside their token file. The repos with genuine call-site leakage are billsplit (16, all ad hoc Colors.red/ teal/black54), epopeja_karta (14, palette inlined in app.dart) and focus_owner (8).
  • There are 4 AppStatusColors, not 5, and they disagree about their own shape: home_inventory and diet-guard have success + warning, habit_stack has success only, wake-alarm has both; home_inventory and habit_stack ship light and dark statics, the other two dark only. The package's version is a superset (success/warning/danger/info/onStatus), so every consumer can adopt it without losing a field.
  • There are 3 section headers, not 5, and no _SectionLabel exists. Two agreed on titleMedium and were reconciled into SectionHeader; epopeja_karta's is a different element — labelSmall + primary color + letter-spacing, i.e. an eyebrow, not a section title. It was deliberately not folded in.

Deferred clusters

Actionable later; each line is repo + path + size, so none of this needs rediscovering.

Flutter — large reconciliations

Cluster Where Size
GitHubMirrorScreen ×5 todo, diet-guard, +3 ~1551 lines, 586 differing between two copies
SettingsScreen ×6 fleet-wide —
List tiles ~20 classes fleet-wide —
Filter sheets todo, dufs-cloud, +1 name collision, not duplicates — dufs-cloud's uses no chips at all; judge structurally
Steppers, badges fleet-wide —

The five GitHubMirrorScreen copies each carry a near-verbatim "kept app-local rather than folded into sync_settings_ui" comment. That decision deserves revisiting on its own, not smuggled into a component extraction.

Flutter — palette drift to correct

Standing decision (2026-08-14): every app uses exactly the frozen palette; where copies disagree, the most popular wins. That is unambiguous — 9 of 10 theme files use the gold #B8862E family.

Repo Drift Fix
testsAndMisc/focus_owner entirely off-palette: cool #5B9DD9 accent, #1B1D21 field, #D9776B danger, #E8EAED text. Its doc comment claims it matches the other com.kuhy.* apps; it does not. Worse, the same six constants are duplicated verbatim in main.dart:16-21 as _kField…_kDanger — the file says they were "lifted out of main.dart", but the originals were never deleted. adopt design_system; delete both copies
kuhylog the only repo using ColorScheme.fromSeed (seed #3B82F6, blue) — the opposite of the convention every other theme file explicitly documents. Also has scoreColor(ColorScheme, int), which is AppStatusColors' job done as a function. adopt design_system + AppStatusColors
epopeja_karta no theme.dart at all; ColorScheme inlined in app.dart:12-20 adopt design_system
billsplit ColorScheme.fromSeed(seedColor: Colors.teal) in main.dart:31, plus 16 ad hoc literals adopt design_system
diet-guard, macro-cam scales drifted from the six-repo consensus adopt design_system

focus_owner is a live enforcement surface, so its visual change wants a human look rather than a blind migration — it is the one item here that is deferred for a reason beyond time.

Remaining Flutter adopters

home_inventory, dufs-cloud, habit_stack, diet-guard, macro-cam, wake-alarm/phone_app, kuhylog, epopeja_karta, billsplit, focus_owner. Each drops its local theme.dart for the tag-pinned dep, as todo and untools now do.

Phase 2 — Python/Tk (~/utils/gatelock), not started

Not a new package: gatelock already owns the tokens (LockConfig), the Tk plumbing and 4 consumers. Add composite widgets, which is where its scope currently stops.

  1. widgets.py::make_button — canonical is diet-guard's _gatelock_buttons.py::make_button, whose variant API picks text color from the fill and so structurally prevents the fg-vs-on_fill contrast bug. leetcode-guard's _button (no <Return> binding) and screen-locker's inline tk.Button are the regressions being retired.
  2. widgets.py::heading() / row() — from leetcode_guard/_status_sections.py.
  3. Export the existing ScrollableSurface and delete leetcode_guard/status_view.py::_scrollable, a hand-rolled reimplementation that lost takefocus and the focus ring. Straight accessibility repair.
  4. widget_group.py::WidgetGroup — the per-output fan-out reimplemented 4× (~916 lines: screen-locker 173, leetcode-guard 105, diet-guard 309, wake-alarm 329). Highest line-count win on this stack.

Proof consumers: diet-guard and leetcode-guard (donor and worst offender). Bump only those two; leave screen-locker and wake-alarm on their current pin — Phase 2 is additive, and restarting live services is blast radius for no gain.

Verification gate — non-negotiable. gatelock backs live systemd services. After any change: /usr/bin/python3 -c "import gatelock" against the real interpreter, not a dev venv. Do not touch the gate window's grab path. Then launch each gate under xvfb-run -s "-screen 0 1366x768x24" and confirm buttons render and <Return> activates them.

Phase 3 — TypeScript (~/utils/web_ui), not started

Lowest confirmed duplication (2 components, 2 repos) and the only stack needing a brand-new consumption mechanism, so it carries the worst effort-to-payoff ratio — sequenced last on purpose.

  1. tokens.css — one :root block. Fixes the worst fork found: 7 repos define 7 different vocabularies for the same palette (--surface-1 vs --panel vs --card; --text vs --ink vs --fg vs --bone; spacing as --space-N vs --sp-N vs --sp-xs). europe-county-map and iron-and-anvil have zero custom properties.
  2. RangeSlider — canonical is dufs-cloud's, whose pure fractionFromPointer(rect, clientX) is testable without layout (jsdom has none). awesome-mcp-explorer's copy has drifted 196 lines.
  3. FilterBar + filter-sort.ts — reconcile dufs-cloud (174) vs awesome-mcp-explorer (225).

Consumption mechanism (decided 2026-08-14): a git dep "@kuhyx/web-ui": "github:kuhyx/utils#web_ui-v0.1.0", matching the Dart/Python tag convention rather than introducing npm workspaces — but shipping a prebuilt dist/ (tsc → ESM + .d.ts) committed in the tag. Raw .tsx in node_modules would need per-repo bundler config, since Vite does not transpile dependencies by default; a committed dist/ keeps consumers at zero config.

Proof consumers: dufs-cloud and awesome-mcp-explorer.

Standing risks

  • Version churn. Each new tag needs a pubspec.yaml bump in every consumer. habit_stack is already 7 minor versions stale on crdt_sync (v0.3.0 vs v0.10.0); the same drift will hit this package.
  • Reconciliation is judgement. Where copies diverged, a "best-of" choice can regress a consumer that depended on its local variant. Migrating proof consumers first exists to surface that early.