[codex] accelerate workspace and editor flows

.github/workflows/performance.yml

@@ -0,0 +1,127 @@
+name: Performance Validation
+
+on:
+  pull_request:
+  push:
+    branches:
+      - main
+  workflow_dispatch:
+
+concurrency:
+  group: performance-validation-${{ github.ref }}
+  cancel-in-progress: true
+
+permissions:
+  contents: read
+
+env:
+  CI: true
+
+jobs:
+  typecheck:
+    name: Typecheck
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm exec tsc --noEmit
+
+  build:
+    name: Build
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm build
+
+  unit:
+    name: Unit Stability
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm exec vitest run tests/save-race.test.ts tests/workspace-race.test.ts
+
+  fidelity:
+    name: Markdown Fidelity
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm exec vitest run tests/fidelity-production.test.ts
+      - run: node scripts/fidelity-gate.mjs docs 10
+
+  perf-smoke:
+    name: Performance Smoke
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm build
+      - name: Check web asset budget
+        run: |
+          node <<'NODE'
+          const fs = require("node:fs");
+          const path = require("node:path");
+          const assetsDir = path.join(process.cwd(), "dist", "assets");
+          const maxAssetBytes = 1_500_000;
+          const assets = fs.existsSync(assetsDir)
+            ? fs.readdirSync(assetsDir).map((name) => path.join(assetsDir, name))
+            : [];
+          const oversized = assets
+            .map((file) => ({ file, size: fs.statSync(file).size }))
+            .filter(({ size }) => size > maxAssetBytes);
+          if (oversized.length > 0) {
+            console.error("Assets above 1.5MB smoke budget:");
+            for (const { file, size } of oversized) {
+              console.error(`${path.relative(process.cwd(), file)}: ${size} bytes`);
+            }
+            process.exit(1);
+          }
+          NODE
+
+  web-e2e:
+    name: Web E2E
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 24
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
+      - run: pnpm exec playwright install --with-deps chromium
+      - run: pnpm test:e2e:web
+
+  # Linux Tauri WebDriver notes:
+  # - Run this only on Linux images with webkit2gtk, xvfb, tauri-driver, and the
+  #   app bundle installed. The web smoke above intentionally does not exercise
+  #   native dialogs, filesystem permissions, or WebView focus behavior.
+  # - A future native lane should start `tauri-driver`, run the packaged app
+  #   under xvfb, then execute WebDriver specs against 127.0.0.1:4444.

docs/performance-validation.md

@@ -0,0 +1,38 @@
+# Performance Validation
+
+This repo validates stability and fidelity in separate CI lanes so failures stay easy to triage.
+
+## CI lanes
+
+- `Typecheck`: `pnpm exec tsc --noEmit`
+- `Build`: `pnpm build`
+- `Unit Stability`: save-race and workspace stale-scan regression tests
+- `Markdown Fidelity`: production editor markdown idempotency tests plus the corpus fidelity gate over `docs`
+- `Performance Smoke`: production build plus a 1.5 MB per-asset smoke budget
+- `Web E2E`: runs headless Playwright against Vite with mocked Tauri APIs
+
+## Local commands
+
+```sh
+pnpm install --frozen-lockfile
+pnpm exec tsc --noEmit
+pnpm build
+pnpm exec vitest run tests/save-race.test.ts tests/workspace-race.test.ts
+pnpm exec vitest run tests/fidelity-production.test.ts
+pnpm exec playwright install chromium
+pnpm test:e2e:web
+pnpm test:ci
+```
+
+## Native Tauri WebDriver
+
+The workflow includes comments for the native Linux WebDriver lane, but does not enable it yet. That lane needs a runner image with `webkit2gtk`, `xvfb`, `tauri-driver`, and the packaged FullMark app available. Once those dependencies are pinned, the lane should:
+
+1. Build the Tauri bundle.
+2. Start `tauri-driver` on `127.0.0.1:4444`.
+3. Launch the packaged app under `xvfb`.
+4. Run WebDriver specs against native filesystem dialogs, focus behavior, and WebView keyboard handling.
+
+The current `web-e2e` lane is intentionally narrower: it proves the frontend
+workflows against mocked Tauri APIs, but it does not validate native filesystem
+dialogs, platform focus behavior, or packaged-app startup.

docs/performance.md

@@ -0,0 +1,71 @@
+# Performance Harness
+
+FullMark's benchmark and fidelity tests use generated vaults only. The harness
+never reads a real user corpus, so CI and local runs are deterministic and safe
+to share.
+
+## Profiles
+
+The generator supports these profiles:
+
+- `small`: 32 markdown notes across a shallow tree.
+- `medium`: 750 notes with frontmatter, wikilinks, and mixed `.md`/`.mdx`.
+- `large`: 5,000 notes across a deeper tree.
+- `chaos`: 1,200 notes plus hidden folders, non-markdown files, long names, and empty directories.
+- `flat-50k`: 50,000 markdown notes in one directory.
+- `deep-5k`: 5,000 notes distributed through a deep tree.
+- `wide-dirs`: 3,000 notes spread across many sibling directories.
+- `mixed-large`: 10,000 notes plus non-markdown and hidden-path noise.
+- `unicode-paths`: 420 notes with Unicode directory and file names.
+
+## Commands
+
+```sh
+npm run test:unit
+npm run test:fidelity
+npm run test:perf
+npm run test:e2e:web
+npm run test:ci
+```
+
+Run every performance profile when you need a full local sweep:
+
+```sh
+node scripts/perf-bench.mjs --profiles all --sample-size 256
+```
+
+The benchmark records these critical paths:
+
+- generated vault creation
+- markdown file discovery
+- sampled `stat` calls
+- sampled file read and frontmatter parse
+- MiniSearch index construction
+- representative query execution
+- sampled TipTap markdown round-trip
+
+## Artifacts
+
+`scripts/perf-bench.mjs` writes JSON and Markdown reports. By default, reports
+and generated vaults are written under the OS temp directory. Set
+`FULLMARK_KEEP_ARTIFACTS=1` to keep generated vaults and reports under `.tmp`:
+
+```sh
+FULLMARK_KEEP_ARTIFACTS=1 npm run test:perf
+```
+
+Set `FULLMARK_ARTIFACT_DIR=/path/to/reports` to choose an explicit report
+directory. Generated vault content is deterministic for a profile and seed.
+
+## Headless web E2E
+
+`npm run test:e2e:web` runs Playwright headlessly against Vite. The tests mock
+Tauri's dialog, event, filesystem, and command APIs before the app loads, then
+exercise the real React UI for workspace open, markdown-only tree filtering,
+source editing and save, quick switching, and reader mode.
+
+Install the browser once on a fresh machine:
+
+```sh
+pnpm exec playwright install chromium
+```

package.json

@@ -2,20 +2,28 @@
   "name": "fullmark",
   "private": true,
   "version": "0.1.5",
+  "packageManager": "pnpm@10.18.2",
   "type": "module",
   "scripts": {
     "dev": "vite",
     "build": "tsc && vite build",
     "build:install": "scripts/build-install.sh",
     "preview": "vite preview",
-    "tauri": "tauri"
+    "tauri": "tauri",
+    "test:unit": "vitest run",
+    "test:fidelity": "node scripts/fidelity-gate.mjs --profiles small,unicode-paths --limit 120",
+    "test:perf": "node scripts/perf-bench.mjs --profiles small,medium,chaos --sample-size 128",
+    "test:e2e:web": "playwright test --config playwright.config.ts",
+    "test:e2e:tauri": "vitest run --passWithNoTests tests/e2e/tauri/**/*.test.ts",
+    "test:ci": "vitest run && node scripts/fidelity-gate.mjs --profiles small,unicode-paths --limit 120 && node scripts/perf-bench.mjs --profiles small --sample-size 32 && pnpm test:e2e:web"
   },
   "dependencies": {
     "@floating-ui/dom": "^1.7.6",
     "@floating-ui/react": "^0.27.19",
     "@tauri-apps/api": "^2",
     "@tauri-apps/plugin-dialog": "^2.7.1",
     "@tauri-apps/plugin-fs": "^2.5.1",
+    "@tauri-apps/plugin-opener": "^2.5.4",
     "@tiptap/core": "^3.23.1",
     "@tiptap/extension-code-block": "^3.23.1",
     "@tiptap/extension-code-block-lowlight": "^3.23.1",
@@ -40,6 +48,7 @@
     "zustand": "^5.0.13"
   },
   "devDependencies": {
+    "@playwright/test": "^1.60.0",
     "@tailwindcss/vite": "^4.3.0",
     "@tauri-apps/cli": "^2",
     "@types/node": "^25.6.2",

playwright.config.ts

@@ -0,0 +1,28 @@
+import { defineConfig, devices } from "@playwright/test";
+
+export default defineConfig({
+  testDir: "tests/e2e/web",
+  timeout: 30_000,
+  expect: {
+    timeout: 5_000,
+  },
+  fullyParallel: true,
+  reporter: process.env.CI ? [["github"], ["list"]] : "list",
+  use: {
+    baseURL: "http://127.0.0.1:1420",
+    trace: "retain-on-failure",
+    screenshot: "only-on-failure",
+  },
+  webServer: {
+    command: "pnpm dev --host 127.0.0.1",
+    url: "http://127.0.0.1:1420",
+    reuseExistingServer: !process.env.CI,
+    timeout: 30_000,
+  },
+  projects: [
+    {
+      name: "chromium",
+      use: { ...devices["Desktop Chrome"] },
+    },
+  ],
+});