perf: use mimalloc as the global allocator

[jdx]

https://entire.io/gh/jdx/hk/trails/211

Summary

Sets mimalloc as hk's global allocator. pklr's config evaluation is allocation-heavy: profiling showed about half its time in malloc/free with the system allocator.

config.rs now spells std::env!("OUT_DIR") explicitly. With the mimalloc crate added, the bare env! became ambiguous with hk's own env module (cannot determine resolution for the macro env), on both stable and beta.

Measurements

hk validate on this repo's hk.pkl, release build, hyperfine with 8 runs:

Allocator	Time
System	1.180 s ± 0.034 s
mimalloc	0.674 s ± 0.021 s

That's 1.75x faster. Separate pklr performance work is in progress (jdx/pklr#223), and the two gains stack.

Testing

• cargo build --release and cargo clippy --all-targets -- -D warnings are clean
• cargo fmt --check is clean
• cargo test: 16 cli::init::detector tests fail on this machine both with and without this change, so they're unrelated

AI-assisted — Tool: Claude Code; model: anthropic/claude-opus-5-5; version: 2.1.286.

🤖 Generated with Claude Code

---

Note

Low Risk
Infrastructure-only change (allocator swap plus a macro path fix); no auth, config semantics, or hook behavior changes.

Overview
Sets mimalloc as hk’s process-wide allocator to speed up allocation-heavy pklr config evaluation (roughly ~1.75× on hk validate in release, per PR benchmarks).

Adds the mimalloc dependency and registers MiMalloc via #[global_allocator] in main.rs.

Fixes a compile-time macro clash: embedded package bytes in config.rs now use std::env!("OUT_DIR") instead of bare env!, because the new dependency made env! ambiguous with hk’s env module.

^{Reviewed by Cursor Bugbot for commit e9d3d83. Bugbot is set up for automated code reviews on this repo. Configure here.}

Summary by CodeRabbit

• Performance
  • Configuration evaluation is now approximately one-third faster, helping the application start more quickly when loading configuration. Actual startup improvement may vary depending on your environment and configuration size.

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	cargo/​mimalloc@​0.1.52

View full report

[coderabbitai]

Navigate logical layers of code changes, visualize relationships, and explore their blast radius.

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

• Configuration used: Repository YAML (base), Central YAML (inherited), Organization UI (inherited)
• Review profile: CHILL
• Plan: Advanced
• Run ID: f1cb0172-b342-4362-9028-c6b9a55d4c0c

📥 Commits

Reviewing files that changed from the base of the PR and between 8e63f1e and e9d3d83.

⛔ Files ignored due to path filters (1)

• Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (3)

• Cargo.toml
• src/config.rs
• src/main.rs

Included review availability: This review used your included allowance. Your plan provides up to 4 included reviews per hour; 3 remain after this review.

---

📝 Walkthrough

Walkthrough

The change adds mimalloc as a dependency and sets it as the global allocator. It also updates the embedded package path to read OUT_DIR with std::env!.

Changes

Runtime and configuration updates

Layer / File(s)	Summary
Add and set the global allocator Cargo.toml, src/main.rs	Cargo.toml adds mimalloc version 0.1. src/main.rs sets mimalloc as the process-wide global allocator.
Read the embedded package path src/config.rs	The embedded package path reads OUT_DIR with std::env! instead of env!.

Priority: ➖ Normal

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Other

Merge Risk: ⚪ Minimal · up to e9d3d

The PR changes hk’s allocator and preserves the generated Pkl package path. No supported-build failure or other actionable merge blocker is established.

Security Architecture Review

Security architecture risk: 🔵 Low · up to e9d3d

The allocator change affects memory management throughout hk, but the inspected changes do not add privileges or new input paths. No concrete security regression was identified. Native dependency behavior and compatibility across release platforms remain incompletely verified.

Retained concerns
No architecture-level concerns identified.

Security review details

Security Blast Radius

• inferred — The directly affected scope is Rust global allocation within the hk process, including configuration evaluation and concurrent runtime work. An allocator defect could therefore affect process-wide memory integrity or availability, although no such defect was identified.

Trust Boundaries and Controls

• inferred — Qualifying std::env! does not turn OUT_DIR into a runtime input. The embedded bytes remain selected during compilation, and the inspected evaluator retains its offline setting, cache configuration, and conditional preload behavior. This change does not expand configuration-input reachability.

Resilience and Maintainability Implications

• inferred — The inspected lifecycle has one fixed global allocator and no newly added manual allocation/free, teardown, or allocator-switch path. Inspected FFI-adjacent code retains local Rust ownership rather than introducing native cleanup of Rust allocations. This counters an ownership-mismatch concern, but does not verify allocation behavior inside transitive native dependencies.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 2 files. (1 skipped: 1…	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely states the main change: using mimalloc as the global allocator.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 2 functions across 2 files. (1 skipped: 1 unsupported.)

• Fix all pre-merge checks with AI

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Replaces the memory allocator with a third-party library.

The PR appears safe to merge; no actionable failure was established.

Summary

This PR makes mimalloc hk’s global allocator to improve allocation-heavy configuration evaluation and qualifies the OUT_DIR macro reference to avoid name ambiguity.

• Adds mimalloc and its native dependency to the manifest and lockfile.
• Applies the allocator unconditionally to the binary.

_{Reviews (1) · Last reviewed commit: "perf: use mimalloc as the global allocat..."}

[github-actions]

Instruction counts

benchmark	trend	instructions	Δ	wall (min)	Δ
builtins	▁▁▁▁▁▁█	1,566,286 → 1,950,643	+24.54% ⚠️	2.40 → 2.93ms	+22.13%
check	—	—	—	16.35 → 17.55ms	+7.39%
config-cold	█████▁	320,480,742 → 257,952,658	-19.51%	59.23 → 49.00ms	-17.27%
stash	█████▁	64,859,813 → 53,518,935	-17.49%	291.05 → 296.51ms	+1.88%
usage	▁▂▃▃▃▃█	4,757,774 → 4,842,879	+1.79% ⚠️	3.55 → 4.71ms	+32.79%
validate	—	—	—	31.18 → 25.56ms	-18.00%

2 benchmark(s) above the 1% gate: builtins +24.54%, usage +1.79%

_{Only instruction counts gate. Wall clock is shown for context — on identical hardware it moves 4-20% run to run.}

_{Measured by tak — instruction-counted CLI benchmarks, stored in this repository's git notes.}

_{e9d3d837e7a1 vs bf1ae647e53e · measured on the runner, not pushed to the history.}