perf(runtime): preserve shared shape index across delete

[proggeramlug]

Summary

• clone the validated slot index when delete forks a shared keys array, while retaining the cheaper move for owned sources
• add an end-to-end regression proving both the untouched sibling and deleting object's shifted clone keep valid accelerators
• add a dedicated 500-key shared-sibling benchmark and changelog fragment; no version metadata is changed

Performance

On bench_shared_shape_delete.ts, 15 interleaved A/B pairs pinned to one CPU:

• median: 171 ms -> 70 ms (-59.1%)
• minimum: 111 ms -> 43 ms (-61.3%)
• checksum: identical in every arm

Both executables were built from commit-matched compiler/runtime pairs on perrymaster.skelpo.net.

Verification

• cargo test -p perry-runtime shared_delete_preserves_the_sibling_shape_index -- --nocapture
• cargo test --quiet -p perry-runtime -- --test-threads=1 (2787 passed, 4 ignored)
• RUSTFLAGS="-D warnings" cargo check -p perry-runtime --all-targets
• scripts/run_lint_gates.sh with compile tier skipped (all 58 runnable lint gates passed; compile covered by the targeted warning-denied check)
• cargo clippy -p perry-runtime --lib

Closes #9006.

Summary by CodeRabbit

• Bug Fixes

  • Improved performance when deleting properties from objects that share internal structures.
  • Preserved fast property lookups on related objects after a deletion.
  • Added regression coverage to ensure sibling objects remain unaffected.

• Benchmarks

  • Added benchmarking for shared-object property deletion and subsequent lookups.
  • Demonstrated a significant reduction in processing time for this scenario.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 53004497-f270-4943-a9ca-4c30a21d5e94

📥 Commits

Reviewing files that changed from the base of the PR and between 5792671 and 0689349.

📒 Files selected for processing (5)

• benchmarks/bench_shared_shape_delete.ts
• changelog.d/9067-preserve-shared-delete-index.md
• crates/perry-runtime/src/object/delete_rest.rs
• crates/perry-runtime/src/object/shapes.rs
• crates/perry-runtime/src/object/shapes_slot_list.rs

Included review availability: Your plan provides up to 8 included reviews per hour; 5 remain after this review.

---

📝 Walkthrough

Walkthrough

The runtime preserves shape-index accelerators for sibling objects when a delete compacts a shared keys array. Owned indexes retain move behavior. A regression test and a benchmark validate sibling index preservation and repeated dynamic lookups.

Changes

Shared delete index preservation

Layer / File(s)	Summary
Runtime index migration crates/perry-runtime/src/object/shapes.rs, crates/perry-runtime/src/object/shapes_slot_list.rs, crates/perry-runtime/src/object/delete_rest.rs	ShapeIndex is cloneable. shape_index_migrate_after_delete clones indexes for shared keys arrays and moves indexes for owned arrays. The delete path passes the shared-state flag.
Regression validation and benchmark crates/perry-runtime/src/object/shapes_slot_list.rs, benchmarks/bench_shared_shape_delete.ts, changelog.d/9067-preserve-shared-delete-index.md	The regression test checks both sibling and deleting-object indexes. The benchmark measures repeated deletes and dynamic sibling lookups. The changelog records the implementation and benchmark results.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Merge Risk: ⚪ Minimal · up to 06893

This change preserves shared shape indexes during property deletion while keeping owned-array deletions efficient, with regression coverage and passing validation checks. No actionable merge-blocking risk remains.

Sequence Diagram(s)

sequenceDiagram
  participant js_object_delete_field
  participant shape_index_migrate_after_delete
  participant sibling_Object_hasOwn
  js_object_delete_field->>shape_index_migrate_after_delete: pass shared keys-array state
  shape_index_migrate_after_delete->>shape_index_migrate_after_delete: clone shared index or move owned index
  sibling_Object_hasOwn->>shape_index_migrate_after_delete: perform indexed sibling lookup

Loading

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the main runtime performance change: preserving a shared shape index across deletion.
Description check	✅ Passed	The description provides a clear summary, concrete changes, linked issue, performance results, and verification commands. It does not use every template heading or complete the checklist, but it conta…
Linked Issues check	✅ Passed	The changes satisfy issue [#9006] by cloning the slot index for shared keys arrays, retaining the move path for owned arrays, adding regression coverage, and providing the requested benchmark.
Out of Scope Changes check	✅ Passed	The implementation, regression test, benchmark, and changelog fragment directly support the linked issue and stated objectives. No unrelated code changes are evident.
Docstring Coverage	✅ Passed	Docstring coverage is 80.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 4 files. (1 skipped: 1 u…

Full details: Description check

Explanation

The description provides a clear summary, concrete changes, linked issue, performance results, and verification commands. It does not use every template heading or complete the checklist, but it contains the required substantive information.

Full details: Docstring Coverage

Explanation

Docstring coverage is 80.00% which is sufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 5 functions across 4 files. (1 skipped: 1 unsupported.)

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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.}

[proggeramlug]

Merged — this closes #9006, which I filed against #9002, and the fix is exactly the right shape: clone for a shared source, move for an owned one.

The asymmetry is the whole point and the comment states it well — forking a shared array is a genuine shape transition (siblings keep naming the old address), while replacing an owned array transfers its identity. get().cloned() vs remove() on that one predicate is a small change for a −59% result.

Verified end-to-end rather than only at the unit level: 25 same-shape siblings of 60 keys with a delete on one — every untouched sibling keeps all 60 keys and correct values, the deleter has 59 with the right hole; ten siblings each deleting a different key; an owned (non-shared) receiver; and delete-then-re-add on one sibling leaving the other untouched. Byte-identical to node, zero sibling corruption.

Worth being precise about what changed, since #9006 was explicitly a throughput issue: correctness was never at risk either way — a stranded sibling rebuilt its index and got the right answer. What this fixes is that it no longer has to.

One process note. Your verification ran run_lint_gates.sh with the compile tier skipped, substituting a targeted cargo check -p perry-runtime --all-targets. I ran the full tier and it's green here, so no harm — but that substitution is exactly what hid the blocker on #9066: perry-runtime compiles standalone with default features and fails under -p perry's feature unification, which a single-crate --all-targets check cannot see. Worth running the compile tier unskipped on anything touching a crate that others depend on.

Validation: the new shared_delete_preserves_the_sibling_shape_index test passes, perry-runtime --lib 2808/0, perry --bins 1066/0, fmt --check, run_lint_gates.sh all 60 gates passed; 2 CI-only skipped (compile tier included).