feat(pack): add optional shared browser runtime

[fireairforce]

Summary

Production browser entries currently embed a separate copy of the runtime. Add optimization.sharedRuntime: true to emit one reusable runtime asset while retaining each entry's bootstrap script. HTML generation and webpack-compatible stats include the runtime in the correct entry asset order. The option defaults to false and does not affect development, Node.js, or library builds; types, schema, documentation, and integration tests are included.

In a local 20-entry build of examples/multi-entries-heavy, the sum of gzip-compressed JavaScript assets decreased by 75.6 KiB (9.9%). A single page loaded one additional script and 110 more gzip bytes, so this is opt-in for applications that benefit from sharing across entries. These measurements do not establish a build-time speedup.

Depends on utooland/next.js#197. This PR advances the submodule to its c037804473bb3ed17b47174258549f12a80854d8 commit; merge the fork prerequisite first. The short-session change in #3395 is independent.

Test Plan

• Native addon and schema generator built with cargo build --locked --profile release-local -p pack-napi -p pack-schema --features plugin; generated schema and compiled both Pack and Pack Shared TypeScript.
• Seven tests passed across sharedRuntime.test.ts, serveStats.test.ts, and serveClientPaths.test.ts, including actual builds, generated HTML ordering, runtime stats, unchanged development/Node output, and Node dynamic import execution.
• Default runtime snapshot passed: cargo test --locked --profile release-local -p pack-tests app_build_runtime -- --nocapture (one matching test; no snapshots updated).
• Local Chrome validation passed all 26 harness commands for the formal configuration: dynamic imports, workers, root/non-root public paths, multiple entries, blocking/deferred HTML scripts, and independent bundles with distinct chunk-loading globals. The browser harness is local validation tooling and is not included in this PR.
• cargo fmt, cargo clippy --all-targets -- -D warnings --no-deps, npx tombi format --check, npx biome ci, typos, and git diff --check passed.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-29T03:41:58.040886Z	ccee183	Draft marked ready

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

📊 Performance Benchmark Report (with-antd)

Utoopack Performance Report

Report ID: utoopack_performance_report_20260929_040050
Generated: 2026-09-29 04:00:50
Trace File: trace_antd.json (0.3GB, 0.82M spans)
Test Project: examples/with-antd

---

Executive Summary

Metric	Value	Assessment
Total Wall Time	6,600.7 ms	Baseline
Total Thread Work (de-duped)	20,471.0 ms	Non-overlapping busy time
Effective Parallelism	3.1x	thread_work / wall_time
Working Threads	10	Threads with actual spans
Thread Utilization	31.0%	⚠️ Suboptimal
Total Spans	824,873	All B/E + X events
Meaningful Spans (>= 10us)	224,746	(27.2% of total)
Tracing Noise (< 10us)	600,127	(72.8% 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	50,153	6,989.4	1,935.9	3,307.9
Parse	8,170	1,354.9	1,016.1	5,908.4
Analyze	145,083	44,392.2	9,502.1	5,825.3
Chunk	5,461	6,029.8	835.3	2,290.3
Codegen	12,525	2,558.1	1,541.3	1,978.3
Emit	32	37.6	18.8	8.0
Other	3,322	7,215.8	4,038.7	6,600.7

---

Workload Distribution by Diagnostic Tier

Category	Spans	Inclusive (ms)	% Work	Self-Time (ms)	% Self
P0: Scheduling & Resolution	195,819	51,896.2	253.5%	11,623.3	56.8%
P1: I/O & Heavy Tasks	2,900	107.9	0.5%	89.1	0.4%
P2: Architecture (Locks/Memory)	0	0.0	0.0%	0.0	0.0%
P3: Asset Pipeline	24,808	9,987.4	48.8%	3,415.2	16.7%
P4: Bridge/Interop	0	0.0	0.0%	0.0	0.0%
Other	1,219	6,586.4	32.2%	3,760.6	18.4%

---

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
4,820.6	26,596.6	97,626	49.4	0.1	11.9	23.5%	module	module (62%)
2,439.4	2,761.0	19	128389.9	327.9	463.9	11.9%	save snapshot	persist (5%)
2,293.5	2,447.1	2,469	928.9	2.8	300.4	11.2%	analyze ecmascript module	module (71%)
1,528.4	14,229.5	34,950	43.7	0.1	6.1	7.5%	process module	process module (82%)
1,253.3	3,358.9	27,375	45.8	0.1	6.0	6.1%	internal resolving	internal resolving (77%)
955.8	1,294.6	6,003	159.2	0.6	40.0	4.7%	parse ecmascript	parse ecmascript (65%)
828.0	915.2	10,227	81.0	0.3	8.7	4.0%	precompute code generation	generate merged code (39%)
724.0	5,729.2	4,019	180.2	0.2	79.7	3.5%	chunking	chunking (47%)
717.9	847.1	7,324	98.0	0.4	102.2	3.5%	compute async module info	compute async module info (57%)
672.7	3,620.6	22,077	30.5	0.0	5.2	3.3%	resolving	module (55%)
625.1	2,057.6	944	662.2	1.6	178.0	3.1%	generate merged code	chunking (45%)
422.0	422.0	329	1282.8	1.1	237.8	2.1%	generate source map	code generation (83%)
345.9	750.9	131	2640.1	5.0	195.7	1.7%	emit code	generate merged code (41%)
322.8	322.8	9	35864.8	162.3	188.6	1.6%	blocking	save snapshot (67%)
291.3	1,220.9	1,969	148.0	0.2	66.8	1.4%	code generation	code generation (83%)
258.9	541.0	1,701	152.2	0.1	129.3	1.3%	write all entrypoints to disk	write all entrypoints to disk (17%)
107.4	296.3	1,371	78.3	0.1	9.4	0.5%	compute async chunks	compute async chunks (47%)
82.9	104.9	819	101.2	0.0	26.8	0.4%	compute binding usage info	compute binding usage info (48%)
60.3	60.3	2,165	27.9	0.0	3.1	0.3%	read file	parse ecmascript (91%)
45.5	80.6	1,870	24.3	0.0	16.1	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	652.5	3	persist → save snapshot → blocking
2	377.3	6	chunking → generate merged code → emit code → emit code → emit code → read file
3	321.8	2	save snapshot → blocking
4	300.8	10	module → module → module → ... → process module → analyze ecmascript module → analyze ecmascript module
5	285.0	3	process module → process module → analyze ecmascript module

---

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	34,950	process module (82%)	43.7 us	0.07 ms	1,528.4 ms
internal resolving	27,375	internal resolving (77%)	45.8 us	0.09 ms	1,253.3 ms

---

Duration Distribution

Range	Count	Percentage
<10us	600,127	72.8%
10us-100us	146,665	17.8%
100us-1ms	66,718	8.1%
1ms-10ms	11,098	1.3%
10ms-100ms	225	0.0%
>100ms	40	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