refactor(pack): use node-stdlib-browser polyfills

[yuzheng14]

Summary

• Replace the copied, ncc-prebundled Node.js polyfills with a generated source dependency closure from node-stdlib-browser@1.3.1.
• Generate and embed 99 aliases across 132 package instances so NAPI and WASM builds do not depend on the consumer's node_modules.
• Preserve the existing 20 unsupported built-ins as empty modules and retain node: protocol aliases.
• Resolve dependencies inside the embedded generated/node_modules boundary before applying the generic foreign-module rule.

Bundle-size comparison

This is a one-off production-build comparison and is intentionally not committed as a snapshot or benchmark fixture.

Methodology

• Both sides use the same @utoo/pack JavaScript wrapper and build configuration; only the NAPI binding changes.
• Configuration: nodePolyfill: true, minification enabled, target: "chrome 100", source maps disabled, persistent caching disabled.
• Every import is retained through a globalThis assignment.
• Each result sums all emitted JavaScript assets. Gzip is level 9; Brotli is quality 11; files are compressed individually before summing.
• The all-modules case imports the 42 canonical module names supported before this migration: 22 real polyfills and 20 legacy empty modules.
• Duplicate node: aliases are not imported because they resolve to the same implementations. Eight new canonical aliases only available after the migration are also excluded to keep the input identical.
• The common-modules case imports buffer, events, path, process, stream, url, and util.
• Values are shown as before → after (change).

Aggregate results

Scenario	Raw	Gzip	Brotli
All 42 existing modules	632.70 → 684.19 KiB (+8.14%)	188.52 → 201.49 KiB (+6.88%)	122.22 → 136.51 KiB (+11.69%)
7 common modules	118.19 → 157.41 KiB (+33.18%)	35.56 → 48.06 KiB (+35.16%)	31.41 → 42.50 KiB (+35.31%)

The 7 common modules are buffer, events, path, process, stream, url, and util.

The embedded NAPI binding changes from 134.28 MiB to 136.51 MiB (+1.66%).

Per-module results

Module	Raw	Gzip	Brotli
assert	66.53 → 75.54 KiB (+13.53%)	20.23 → 22.66 KiB (+12.03%)	18.09 → 20.28 KiB (+12.10%)
async_hooks (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
buffer	33.04 → 33.10 KiB (+0.18%)	11.07 → 10.99 KiB (-0.73%)	9.93 → 9.84 KiB (-0.83%)
child_process (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
cluster (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
constants	14.36 → 14.36 KiB (0.00%)	5.80 → 5.80 KiB (0.00%)	5.14 → 5.14 KiB (+0.02%)
crypto	447.04 → 471.44 KiB (+5.46%)	133.18 → 139.80 KiB (+4.97%)	91.95 → 92.81 KiB (+0.94%)
dgram (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
diagnostics_channel (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
dns (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
domain	18.08 → 17.53 KiB (-3.06%)	6.52 → 6.32 KiB (-3.03%)	5.86 → 5.67 KiB (-3.20%)
events	17.10 → 16.86 KiB (-1.44%)	6.25 → 6.13 KiB (-1.97%)	5.63 → 5.49 KiB (-2.40%)
fs (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
fs/promises (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
http	161.22 → 138.04 KiB (-14.37%)	47.82 → 42.66 KiB (-10.79%)	32.91 → 37.92 KiB (+15.23%)
http2 (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
https	161.96 → 138.45 KiB (-14.51%)	47.98 → 42.79 KiB (-10.83%)	33.00 → 38.09 KiB (+15.41%)
inspector (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
module (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
net (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
os	11.77 → 11.65 KiB (-0.96%)	4.53 → 4.46 KiB (-1.44%)	4.00 → 3.95 KiB (-1.32%)
path	15.33 → 15.09 KiB (-1.59%)	5.94 → 5.82 KiB (-1.99%)	5.33 → 5.21 KiB (-2.18%)
perf_hooks (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
process	12.86 → 14.22 KiB (+10.51%)	4.98 → 5.34 KiB (+7.32%)	4.42 → 4.77 KiB (+7.86%)
punycode	13.48 → 13.57 KiB (+0.70%)	5.46 → 5.54 KiB (+1.45%)	4.89 → 4.95 KiB (+1.26%)
querystring	12.71 → 13.13 KiB (+3.29%)	4.98 → 5.02 KiB (+0.84%)	4.45 → 4.49 KiB (+0.90%)
readline (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
repl (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
setimmediate	12.71 → 12.58 KiB (-1.01%)	5.02 → 4.95 KiB (-1.40%)	4.43 → 4.35 KiB (-1.72%)
stream	106.21 → 78.20 KiB (-26.37%)	31.45 → 23.34 KiB (-25.80%)	27.79 → 20.93 KiB (-24.70%)
string_decoder	37.53 → 37.31 KiB (-0.58%)	12.12 → 12.01 KiB (-0.89%)	10.88 → 10.75 KiB (-1.25%)
timers	13.96 → 13.53 KiB (-3.13%)	5.40 → 5.21 KiB (-3.51%)	4.78 → 4.61 KiB (-3.53%)
tls (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
trace_events (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
tty	11.34 → 11.22 KiB (-1.02%)	4.39 → 4.32 KiB (-1.54%)	3.89 → 3.83 KiB (-1.43%)
url	16.81 → 64.11 KiB (+281.33%)	6.65 → 21.18 KiB (+218.55%)	6.01 → 19.08 KiB (+217.16%)
util	38.21 → 42.85 KiB (+12.16%)	12.29 → 13.74 KiB (+11.83%)	11.03 → 12.36 KiB (+12.05%)
v8 (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
vm	13.59 → 13.30 KiB (-2.12%)	5.23 → 5.10 KiB (-2.32%)	4.62 → 4.51 KiB (-2.28%)
wasi (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
worker_threads (empty)	11.06 → 11.06 KiB (0.00%)	4.25 → 4.25 KiB (0.00%)	3.76 → 3.76 KiB (0.00%)
zlib	211.72 → 191.71 KiB (-9.45%)	61.52 → 56.14 KiB (-8.74%)	47.52 → 48.96 KiB (+3.01%)

Observations

• By Gzip size, 14 modules become smaller, 7 become larger, and 21 are unchanged (the 20 legacy empty modules plus constants).
• stream has the largest reduction: 31.45 → 23.34 KiB (-25.80% Gzip).
• url has the largest increase: 6.65 → 21.18 KiB (+218.55% Gzip), because the previous compact native-url implementation is replaced by the fuller node-stdlib-browser URL compatibility implementation and its dependencies.
• The complete 42-module bundle grows by 6.88% Gzip; the common-module bundle grows by 35.16% Gzip.

Test Plan

• cargo fmt --check
• cargo clippy --all-targets -- -D warnings --no-deps
• cargo test -p pack-core node_polyfill
• cargo test -p pack-tests --test snapshot node_polyfill
• cargo test -p pack-tests (140 passed, 37 ignored)
• npm run build-wasm --workspace @utoo/web -- --profile release-local
• Regenerated the polyfill manifest and dependency closure with npm run gen:node-polyfills from crates/pack-core/js.

Closes #2542

[github-actions]

📊 Performance Benchmark Report (with-antd)

Utoopack Performance Report

Report ID: utoopack_performance_report_20260914_083505
Generated: 2026-09-14 08:35:05
Trace File: trace_antd.json (0.3GB, 0.82M spans)
Test Project: examples/with-antd

---

Executive Summary

Metric	Value	Assessment
Total Wall Time	5,284.2 ms	Baseline
Total Thread Work (de-duped)	15,787.9 ms	Non-overlapping busy time
Effective Parallelism	3.0x	thread_work / wall_time
Working Threads	10	Threads with actual spans
Thread Utilization	29.9%	⚠️ Suboptimal
Total Spans	820,006	All B/E + X events
Meaningful Spans (>= 10us)	177,618	(21.7% of total)
Tracing Noise (< 10us)	642,388	(78.3% of total)

---

Build Phase Timeline

Shows when each build phase is active and how much CPU it consumes.
Self-Time is the time spent exclusively in that phase (excluding children).

Phase	Spans	Inclusive (ms)	Self-Time (ms)	Wall Range (ms)
Resolve	38,375	4,674.7	1,237.3	2,338.4
Parse	8,150	1,028.2	742.2	4,586.6
Analyze	114,819	31,668.4	6,551.4	4,513.4
Chunk	5,178	4,825.1	802.3	1,947.5
Codegen	8,829	1,841.2	1,147.5	1,699.1
Emit	29	32.2	16.1	7.8
Other	2,238	6,340.3	3,821.0	5,284.2

---

Workload Distribution by Diagnostic Tier

Category	Spans	Inclusive (ms)	% Work	Self-Time (ms)	% Self
P0: Scheduling & Resolution	153,476	36,754.4	232.8%	7,977.5	50.5%
P1: I/O & Heavy Tasks	2,701	92.9	0.6%	76.8	0.5%
P2: Architecture (Locks/Memory)	0	0.0	0.0%	0.0	0.0%
P3: Asset Pipeline	20,637	7,717.7	48.9%	2,703.4	17.1%
P4: Bridge/Interop	0	0.0	0.0%	0.0	0.0%
Other	804	5,845.2	37.0%	3,560.0	22.5%

---

Top 20 Tasks by Self-Time

Self-time is the exclusive duration: time spent in the task itself, not in sub-tasks.
This is the most accurate indicator of where CPU cycles are actually spent.

Self (ms)	Inclusive (ms)	Count	Avg Self (us)	P95 Self (ms)	Max Self (ms)	% Work	Task Name	Top Caller
3,178.1	19,064.2	79,192	40.1	0.1	8.4	20.1%	module	module (60%)
2,594.5	2,693.5	16	162159.0	358.8	577.1	16.4%	save snapshot	persist (6%)
1,772.1	1,857.7	2,152	823.4	2.2	312.3	11.2%	analyze ecmascript module	module (81%)
961.5	9,950.1	25,917	37.1	0.1	5.8	6.1%	process module	process module (79%)
811.3	2,220.6	21,075	38.5	0.1	4.1	5.1%	internal resolving	internal resolving (71%)
708.1	4,668.6	3,904	181.4	0.2	68.3	4.5%	chunking	chunking (49%)
686.6	972.6	5,981	114.8	0.4	28.9	4.3%	parse ecmascript	parse ecmascript (65%)
629.7	699.8	6,753	93.2	0.4	6.1	4.0%	precompute code generation	generate merged code (48%)
535.6	609.0	5,359	99.9	0.3	92.0	3.4%	compute async module info	compute async module info (49%)
534.0	1,714.2	571	935.2	2.4	216.9	3.4%	generate merged code	chunking (78%)
420.8	2,449.0	16,797	25.1	0.0	4.3	2.7%	resolving	module (62%)
351.1	351.1	329	1067.1	0.8	204.5	2.2%	generate source map	code generation (83%)
312.6	642.2	131	2385.9	4.8	182.6	2.0%	emit code	generate merged code (41%)
247.6	423.9	1,091	226.9	0.1	144.2	1.6%	write all entrypoints to disk	write all entrypoints to disk (8%)
166.7	790.4	1,747	95.4	0.2	35.2	1.1%	code generation	code generation (82%)
100.0	100.0	8	12495.5	49.8	62.4	0.6%	blocking	save snapshot (62%)
91.3	153.2	1,223	74.6	0.1	8.5	0.6%	compute async chunks	compute async chunks (57%)
67.0	78.8	649	103.3	0.0	26.6	0.4%	compute binding usage info	compute binding usage info (55%)
55.6	55.6	2,166	25.6	0.0	3.8	0.4%	read file	parse ecmascript (91%)
25.9	49.0	1,533	16.9	0.0	10.3	0.2%	collect mergeable modules	collect mergeable modules (100%)

---

Critical Path Analysis

The longest sequential dependency chains that determine wall-clock time.
Focus on reducing the depth of these chains to improve parallelism.

Rank	Self-Time (ms)	Depth	Path
1	639.5	3	persist → save snapshot → blocking
2	402.4	6	chunking → generate merged code → emit code → emit code → emit code → read file
3	312.5	7	module → module → process module → process module → process module → process module → analyze ecmascript module
4	282.7	9	module → module → module → ... → process module → analyze ecmascript module → analyze ecmascript module
5	237.1	4	chunking → generate merged code → emit code → generate source map

---

Batching Candidates

High-volume tasks dominated by a single parent. If the parent can batch them,
it drastically reduces scheduler overhead.

Task Name	Count	Top Caller (Attribution)	Avg Self	P95 Self	Total Self
process module	25,917	process module (79%)	37.1 us	0.05 ms	961.5 ms
internal resolving	21,075	internal resolving (71%)	38.5 us	0.06 ms	811.3 ms

---

Duration Distribution

Range	Count	Percentage
<10us	642,388	78.3%
10us-100us	113,389	13.8%
100us-1ms	56,948	6.9%
1ms-10ms	7,110	0.9%
10ms-100ms	128	0.0%
>100ms	43	0.0%

---

Action Items

1. [P0] Focus on tasks with the highest Self-Time — these are where CPU cycles are actually spent.
2. [P0] Use Batching Candidates to identify callers that should use try_join or reduce #[turbo_tasks::function] granularity.
3. [P1] Check Build Phase Timeline for phases with disproportionate wall range vs. self-time (= serialization).
4. [P1] Inspect P95 Self (ms) for heavy monolith tasks. Focus on long-tail outliers, not averages.
5. [P1] Review Critical Paths — reducing the longest chain depth directly improves wall-clock time.
6. [P2] If Thread Utilization < 60%, investigate scheduling gaps (lock contention or deep dependency chains).

Report generated by Utoopack Performance Analysis Agent