feat: optimize react testing runtime and add prettier

.prettierrc.json

@@ -0,0 +1,34 @@
+{
+  "printWidth": 80,
+  "tabWidth": 2,
+  "semi": true,
+  "singleQuote": true,
+  "quoteProps": "as-needed",
+  "jsxSingleQuote": true,
+  "trailingComma": "es5",
+  "bracketSpacing": true,
+  "bracketSameLine": false,
+  "arrowParens": "always",
+  "proseWrap": "preserve",
+  "htmlWhitespaceSensitivity": "css",
+  "endOfLine": "auto",
+  "embeddedLanguageFormatting": "auto",
+  "singleAttributePerLine": false,
+  "plugins": ["@ianvs/prettier-plugin-sort-imports"],
+  "importOrder": [
+    "<TYPES>^(node:)",
+    "<TYPES>",
+    "<TYPES>^[.]",
+    "<BUILTIN_MODULES>",
+    "<THIRD_PARTY_MODULES>",
+    "^[.]"
+  ],
+  "overrides": [
+    {
+      "files": "*.jsonc",
+      "options": {
+        "trailingComma": "none"
+      }
+    }
+  ]
+}

CHANGELOG.md

@@ -9,7 +9,6 @@
 
 ## 1.0.0 (2026-03-31)
 
-
 ### Bug Fixes
 
-* clarify Deno compatibility guidance in README ([#2](https://github.com/Lojhan/poku-react-testing/issues/2)) ([ee28792](https://github.com/Lojhan/poku-react-testing/commit/ee28792965b9cc9dacc8bee2fc96a5f84088f2d1))
+- clarify Deno compatibility guidance in README ([#2](https://github.com/Lojhan/poku-react-testing/issues/2)) ([ee28792](https://github.com/Lojhan/poku-react-testing/commit/ee28792965b9cc9dacc8bee2fc96a5f84088f2d1))

README.md

@@ -115,11 +115,11 @@ test('renders a heading', () => {
 
 ### Runtime × DOM Adapter
 
-| | Node.js ≥ 20 | Bun ≥ 1 | Deno ≥ 2 |
-|---|:---:|:---:|:---:|
-| **happy-dom** | ✅ | ✅ | ✅ |
-| **jsdom** | ✅ | ✅ | ⚠️ |
-| **custom setup** | ✅ | ✅ | ✅ |
+|                  | Node.js ≥ 20 | Bun ≥ 1 | Deno ≥ 2 |
+| ---------------- | :----------: | :-----: | :------: |
+| **happy-dom**    |      ✅      |   ✅    |    ✅    |
+| **jsdom**        |      ✅      |   ✅    |    ⚠️    |
+| **custom setup** |      ✅      |   ✅    |    ✅    |
 
 > [!NOTE]
 >

benchmark/REPORT.md

@@ -0,0 +1,124 @@
+# React Testing Framework Benchmark Report
+
+> Generated: Tue, 31 Mar 2026 21:00:50 GMT
+
+## Environment
+
+| Property | Value |
+|---|---|
+| Node.js | v22.5.1 |
+| Platform | darwin 25.4.0 |
+| CPU | Apple M3 Pro |
+| CPU Cores | 12 |
+| Total RAM | 18.0 GB |
+| Runs/scenario | 7 (trim ±1) |
+
+## Scenarios
+
+Each scenario runs the **same 9 React tests** across 5 test files:
+
+| Test File | Tests |
+|---|---|
+| 'counter.test.jsx' | 1 — stateful counter, event interaction |
+| 'hooks.test.jsx' | 2 — custom hook harness + `renderHook` |
+| 'lifecycle.test.jsx' | 2 — `rerender`, `unmount` + effect cleanup |
+| 'context.test.jsx' | 1 — `createContext` + wrapper injection |
+| 'concurrency.test.jsx' | 2 — React 19 `use()` + `useTransition` |
+
+### Frameworks under test
+
+| Combination | DOM layer | Assertion style |
+|---|---|---|
+| poku + poku-react-testing | happy-dom | `assert.strictEqual` |
+| poku + poku-react-testing | jsdom | `assert.strictEqual` |
+| jest 29 + @testing-library/react | jsdom (jest-environment-jsdom) | `expect().toBe()` |
+| vitest 3 + @testing-library/react | jsdom | `expect().toBe()` |
+| vitest 3 + @testing-library/react | happy-dom | `expect().toBe()` |
+
+## Results
+
+| Scenario           | Mean   | Min    | Max    | Stdev  | Peak RSS | vs poku+happy-dom |
+|--------------------|--------|--------|--------|--------|----------|-------------------|
+| poku + happy-dom   | 1.073s | 0.996s | 1.230s | 0.085s | 163.3 MB | *(baseline)*      |
+| poku + jsdom       | 1.060s | 1.007s | 1.177s | 0.060s | 163.4 MB | -1%               |
+| jest + jsdom       | 0.859s | 0.779s | 0.929s | 0.050s | 206.2 MB | -20%              |
+| vitest + jsdom     | 0.964s | 0.950s | 0.987s | 0.017s | 148.0 MB | -10%              |
+| vitest + happy-dom | 0.838s | 0.812s | 0.864s | 0.021s | 116.3 MB | -22%              |
+
+> **Wall-clock time** is measured with `performance.now()` around the child-process spawn.
+> **Peak RSS** is captured via `/usr/bin/time -l` on macOS (bytes → MB).
+> The baseline for relative comparisons is **poku + happy-dom**.
+
+## Analysis
+
+### Overall ranking (mean wall-clock time)
+
+1. **vitest + happy-dom** — 0.838s
+2. **jest + jsdom** — 0.859s
+3. **vitest + jsdom** — 0.964s
+4. **poku + jsdom** — 1.060s
+5. **poku + happy-dom** — 1.073s
+
+### Speed comparison
+
+- poku+happy-dom vs jest+jsdom: jest is **-20% faster**
+- poku+happy-dom vs vitest+jsdom: vitest is **-10% faster**
+- jest+jsdom vs vitest+jsdom: vitest is **12% slower** than jest
+
+### DOM adapter impact
+
+- **poku**: happy-dom vs jsdom — jsdom is **-1% faster**
+- **vitest**: happy-dom vs jsdom — jsdom is **15% slower**
+
+### Memory footprint
+
+- **vitest + happy-dom**: 116.3 MB peak RSS
+- **vitest + jsdom**: 148.0 MB peak RSS
+- **poku + happy-dom**: 163.3 MB peak RSS
+- **poku + jsdom**: 163.4 MB peak RSS
+- **jest + jsdom**: 206.2 MB peak RSS
+
+### Consistency (lower stdev = more predictable)
+
+- **vitest + jsdom**: σ = 0.017s
+- **vitest + happy-dom**: σ = 0.021s
+- **jest + jsdom**: σ = 0.050s
+- **poku + jsdom**: σ = 0.060s
+- **poku + happy-dom**: σ = 0.085s
+
+## Key findings
+
+- **Fastest**: vitest + happy-dom — 0.838s mean
+- **Slowest**: poku + happy-dom — 1.073s mean
+- **Speed spread**: 28% difference between fastest and slowest
+
+### Interpretation
+
+**poku + poku-react-testing** avoids the multi-process or bundler startup that jest (babel transform
+pipeline) and vitest (Vite + module graph) require. Its architecture — isolated per-file Node.js
+processes with minimal bootstrap — means cold-start overhead is proportional to the number of test
+files, not to the framework's own initialization.
+
+**jest** carries the heaviest startup cost due to:
+1. Babel transformation of every TSX file on first run (no persistent cache in this benchmark)
+2. 'jest-worker' process pool initialisation
+3. JSDOM environment setup per test file
+
+**vitest** starts faster than jest because Vite's module graph is more efficient, and the
+esbuild/Rollup pipeline is faster than Babel. However, the Vite dev server and HMR machinery still
+contribute to startup overhead compared to a zero-bundler approach.
+
+**DOM adapter choice** (happy-dom vs jsdom) has a measurable but smaller effect than the choice of
+framework. happy-dom is generally lighter and initialises faster; jsdom is more spec-complete.
+
+## Reproducibility
+
+```sh
+# Install benchmark deps (one-time)
+cd benchmark && npm install && cd ..
+
+# Re-run with custom run count
+BENCH_RUNS=10 node benchmark/run.mjs
+```
+
+Results are saved to `benchmark/results.json` for programmatic analysis.

benchmark/babel.config.cjs

@@ -0,0 +1,6 @@
+module.exports = {
+  presets: [
+    ['@babel/preset-env', { targets: { node: 'current' } }],
+    ['@babel/preset-react', { runtime: 'automatic' }],
+  ],
+};

benchmark/jest.config.mjs

@@ -0,0 +1,8 @@
+export default {
+  testEnvironment: 'jsdom',
+  transform: {
+    '^.+\\.[jt]sx?$': ['babel-jest', { configFile: './babel.config.cjs' }],
+  },
+  testMatch: ['<rootDir>/tests/jest/**/*.test.jsx'],
+  moduleFileExtensions: ['js', 'jsx', 'json'],
+};