feat: extract shared DOM/testing core to @pokujs/dom + benchmark suite

benchmark/REPORT.md

@@ -1,96 +1,97 @@
 # React Testing Framework Benchmark Report
 
-> Generated: Wed, 01 Apr 2026 12:41:50 GMT
+> Generated: Sun, 05 Apr 2026 19:06:36 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)   |
+| 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`     |
+| 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 + @pokujs/react              | happy-dom                      | `assert.strictEqual` |
-| poku + @pokujs/react              | 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()`    |
+| Combination | DOM layer | Assertion style |
+|---|---|---|
+| poku + @pokujs/react | happy-dom | `assert.strictEqual` |
+| poku + @pokujs/react | 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   | 0.560s | 0.515s | 0.600s | 0.033s | 154.3 MB | _(baseline)_      |
-| poku + jsdom       | 0.444s | 0.429s | 0.451s | 0.008s | 157.1 MB | -21%              |
-| jest + jsdom       | 1.040s | 0.975s | 1.135s | 0.056s | 203.4 MB | +86%              |
-| vitest + jsdom     | 1.193s | 1.129s | 1.269s | 0.057s | 152.3 MB | +113%             |
-| vitest + happy-dom | 1.041s | 0.990s | 1.126s | 0.047s | 117.1 MB | +86%              |
-
-> **Wall-clock time** is measured with `performance.now()` around the child-process spawn.
+|--------------------|--------|--------|--------|--------|----------|-------------------|
+| poku + happy-dom   | 0.115s | 0.112s | 0.109s | 0.004s | 127.4 MB | *(baseline)*      |
+| poku + jsdom       | 0.294s | 0.280s | 0.299s | 0.007s | 172.1 MB | +157%             |
+| jest + jsdom       | 0.843s | 0.768s | 0.879s | 0.041s | 198.8 MB | +636%             |
+| vitest + jsdom     | 0.968s | 0.951s | 0.986s | 0.011s | 147.8 MB | +745%             |
+| vitest + happy-dom | 0.845s | 0.825s | 0.856s | 0.013s | 115.2 MB | +637%             |
+
+> **Poku elapsed time** uses Poku's reported suite `Duration` (ANSI-stripped parse) to avoid teardown-skew artifacts.
+> **Jest/Vitest elapsed 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. **poku + jsdom** — 0.444s
-2. **poku + happy-dom** — 0.560s
-3. **jest + jsdom** — 1.040s
-4. **vitest + happy-dom** — 1.041s
-5. **vitest + jsdom** — 1.193s
+1. **poku + happy-dom** — 0.115s
+2. **poku + jsdom** — 0.294s
+3. **jest + jsdom** — 0.843s
+4. **vitest + happy-dom** — 0.845s
+5. **vitest + jsdom** — 0.968s
 
 ### Speed comparison
 
-- poku+happy-dom vs jest+jsdom: jest is **86% slower**
-- poku+happy-dom vs vitest+jsdom: vitest is **113% slower**
+- poku+happy-dom vs jest+jsdom: jest is **636% slower**
+- poku+happy-dom vs vitest+jsdom: vitest is **745% slower**
 - jest+jsdom vs vitest+jsdom: vitest is **15% slower** than jest
 
 ### DOM adapter impact
 
-- **poku**: happy-dom vs jsdom — jsdom is **-21% faster**
+- **poku**: happy-dom vs jsdom — jsdom is **157% slower**
 - **vitest**: happy-dom vs jsdom — jsdom is **15% slower**
 
 ### Memory footprint
 
-- **vitest + happy-dom**: 117.1 MB peak RSS
-- **vitest + jsdom**: 152.3 MB peak RSS
-- **poku + happy-dom**: 154.3 MB peak RSS
-- **poku + jsdom**: 157.1 MB peak RSS
-- **jest + jsdom**: 203.4 MB peak RSS
+- **vitest + happy-dom**: 115.2 MB peak RSS
+- **poku + happy-dom**: 127.4 MB peak RSS
+- **vitest + jsdom**: 147.8 MB peak RSS
+- **poku + jsdom**: 172.1 MB peak RSS
+- **jest + jsdom**: 198.8 MB peak RSS
 
 ### Consistency (lower stdev = more predictable)
 
-- **poku + jsdom**: σ = 0.008s
-- **poku + happy-dom**: σ = 0.033s
-- **vitest + happy-dom**: σ = 0.047s
-- **jest + jsdom**: σ = 0.056s
-- **vitest + jsdom**: σ = 0.057s
+- **poku + happy-dom**: σ = 0.004s
+- **poku + jsdom**: σ = 0.007s
+- **vitest + jsdom**: σ = 0.011s
+- **vitest + happy-dom**: σ = 0.013s
+- **jest + jsdom**: σ = 0.041s
 
 ## Key findings
 
-- **Fastest**: poku + jsdom — 0.444s mean
-- **Slowest**: vitest + jsdom — 1.193s mean
-- **Speed spread**: 169% difference between fastest and slowest
+- **Fastest**: poku + happy-dom — 0.115s mean
+- **Slowest**: vitest + jsdom — 0.968s mean
+- **Speed spread**: 745% difference between fastest and slowest
 
 ### Interpretation
 
@@ -100,7 +101,6 @@ processes with minimal bootstrap — means cold-start overhead is proportional t
 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

benchmark/results.json

@@ -1,5 +1,5 @@
 {
-  "timestamp": "2026-04-01T12:41:50.774Z",
+  "timestamp": "2026-04-05T19:06:36.388Z",
   "system": {
     "nodeVersion": "v22.5.1",
     "platform": "darwin 25.4.0",
@@ -14,68 +14,68 @@
   "results": [
     {
       "label": "poku + happy-dom",
-      "mean": 559.9537916000002,
-      "min": 515.060708,
-      "max": 599.9880840000001,
-      "stddev": 33.06234198220535,
-      "meanRss": 161808384,
-      "meanUserCpu": 672,
-      "meanSysCpu": 202,
+      "mean": 114.5411498,
+      "min": 111.865375,
+      "max": 108.566625,
+      "stddev": 4.164501573153617,
+      "meanRss": 133624627.2,
+      "meanUserCpu": 358,
+      "meanSysCpu": 80,
       "runs": 7,
       "validRuns": 7,
       "failures": 0
     },
     {
       "label": "poku + jsdom",
-      "mean": 443.6319584000001,
-      "min": 428.5047089999998,
-      "max": 450.59829100000024,
-      "stddev": 7.9857833050707905,
-      "meanRss": 164767334.4,
-      "meanUserCpu": 432,
-      "meanSysCpu": 92,
+      "mean": 294.0646082,
+      "min": 280.499125,
+      "max": 298.546375,
+      "stddev": 6.855649968732701,
+      "meanRss": 180456652.8,
+      "meanUserCpu": 532,
+      "meanSysCpu": 102,
       "runs": 7,
       "validRuns": 7,
       "failures": 0
     },
     {
       "label": "jest + jsdom",
-      "mean": 1040.1169332000002,
-      "min": 975.1540420000001,
-      "max": 1134.9154159999998,
-      "stddev": 56.06532298235639,
-      "meanRss": 213300019.2,
-      "meanUserCpu": 936,
-      "meanSysCpu": 228,
+      "mean": 842.5360584000006,
+      "min": 768.3345829999998,
+      "max": 879.3446250000015,
+      "stddev": 40.57889525259689,
+      "meanRss": 208506060.8,
+      "meanUserCpu": 784,
+      "meanSysCpu": 152,
       "runs": 7,
       "validRuns": 7,
       "failures": 0
     },
     {
       "label": "vitest + jsdom",
-      "mean": 1192.7835997999996,
-      "min": 1129.2755830000006,
-      "max": 1268.7202919999982,
-      "stddev": 56.83080807294834,
-      "meanRss": 159652249.6,
-      "meanUserCpu": 3798,
-      "meanSysCpu": 1380,
+      "mean": 967.6274329999997,
+      "min": 950.946915999999,
+      "max": 985.529625000001,
+      "stddev": 11.190019303221952,
+      "meanRss": 155002470.4,
+      "meanUserCpu": 3524,
+      "meanSysCpu": 954,
       "runs": 7,
       "validRuns": 7,
       "failures": 0
     },
     {
       "label": "vitest + happy-dom",
-      "mean": 1041.3667920000007,
-      "min": 990.0950420000008,
-      "max": 1126.3293750000012,
-      "stddev": 47.26983715919258,
-      "meanRss": 122814464,
-      "meanUserCpu": 3346,
-      "meanSysCpu": 1156,
+      "mean": 844.6526001999998,
+      "min": 824.9157500000001,
+      "max": 856.1071249999986,
+      "stddev": 13.21280400724632,
+      "meanRss": 120832000,
+      "meanUserCpu": 3056,
+      "meanSysCpu": 900,
       "runs": 7,
       "validRuns": 7,
       "failures": 0
     }
   ]
-}
+}