perf(codegen): speed up generic registry pipelines

[proggeramlug]

Summary

Speeds up the pipeline benchmark by keeping stable array-valued fields in local slots for methods called on proven-contained receivers, and by adding a guarded numeric fast path for null-defaulted dynamic counters. Dynamic JavaScript behavior and aliased receiver routes retain their existing fallbacks.

Changes

• Emit a separate internal $ptr_arrays method clone for eligible loop methods on Phase 3b contained receivers.
• Snapshot only array fields initialized by array literals, and decline the clone for slot replacement, deletion, or reentrant this/super method calls.
• Keep aliased, guarded, and dispatch-tower calls on the existing public/$pshape paths.
• Recognize the (value === null ? 0 : value) + numeric-literal shape and use an inline numeric add after a runtime tag check, with full dynamic add as the fallback.
• Add IR routing/safety tests and an end-to-end parity regression.

Related issue

Closes #8607

Test plan

• cargo build --release -p perry
• cargo test -p perry-codegen --lib --quiet (1,163 passed; 1 ignored)
• PERRY_SKIP_BUILD=1 ./run_parity_tests.sh --filter test_issue_8607 (1/1 passed, byte-exact against Node)
• python3 scripts/check_test_registration.py
• cargo fmt --all -- --check
• ./scripts/pre-tag-check.sh --quick
• Full workspace test command from the template. The affected-crate runner completed all 2,634 runtime tests (4 ignored), then stopped in an unchanged perry binary audit because current main does not classify the PERRY_OUTLINE_ENTRY environment variables in the build-cache allow/exclusion lists.

The private m0810 20-row acceptance corpus referenced by repository notes is not present in this checkout; the public pipeline workload and the added parity fixture are byte-exact.

Benchmark

Apple M1 mini, quiet host. Wall spread is 20 shuffled/interleaved five-process batches, normalized per process; batching avoids the 10 ms resolution of macOS /usr/bin/time. Counters are 20 shuffled/interleaved single-process runs. Baseline is 970eab3; final is this branch.

runtime	wall median	wall range	instructions median	cycles median	peak RSS
baseline	114.101 ms	113.263-115.746 ms	2,131,456,485	348,841,468	22,282,240 B
final	92.976 ms	92.229-95.320 ms	1,784,508,872	283,436,554	22,282,240 B
Node	101.476 ms	99.407-104.134 ms	1,231,885,256	355,256,028	90,112,000 B

Final versus baseline: 18.5% lower wall time with disjoint ranges, 16.3% fewer instructions, 18.7% fewer cycles, and unchanged peak RSS. Final is 8.4% faster than Node by median wall time on this host.

Screenshots / output

pipeline output from both Perry and Node:

55626000 3 3

Checklist

• I have NOT bumped the workspace version or edited CLAUDE.md / CHANGELOG.md
• My commit follows the repository commit-prefix convention
• I have read CONTRIBUTING.md and agree to the Code of Conduct

Summary by CodeRabbit

• Performance Improvements

  • Improved performance for methods that repeatedly access cached array fields on contained receivers.
  • Added optimized numeric addition for null-defaulted counters while preserving dynamic string-or-number behavior.

• Bug Fixes

  • Preserved correct behavior for aliased receivers, non-numeric fallback values, and property rebinding scenarios.

• Tests

  • Added coverage for cached array-field routing and null-defaulted numeric addition, including missing, repeated, and string-valued entries.

[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: 36df0289-454c-45e8-bb0c-e21581c0b573

📥 Commits

Reviewing files that changed from the base of the PR and between 5f2c6c6 and 68c8eb4.

📒 Files selected for processing (11)

• changelog.d/8607-pipeline-null-default-add.md
• crates/perry-codegen/src/codegen/artifacts.rs
• crates/perry-codegen/src/codegen/method.rs
• crates/perry-codegen/src/collectors/mod.rs
• crates/perry-codegen/src/collectors/proven_this.rs
• crates/perry-codegen/src/collectors/proven_this_routing_tests.rs
• crates/perry-codegen/src/expr/binary.rs
• crates/perry-codegen/src/expr/mod.rs
• crates/perry-codegen/src/expr/null_default_numeric_add_tests.rs
• crates/perry-codegen/src/lower_call/property_get/dynamic_dispatch.rs
• test-files/test_issue_8607_null_default_numeric_add.ts

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

---

📝 Walkthrough

Walkthrough

The change adds guarded numeric lowering for null-defaulted additions and array-field caching for eligible contained receivers. It updates clone compilation and dispatch routing, and adds IR, routing, runtime, and changelog coverage.

Changes

Pipeline optimizations

Layer / File(s)	Summary
Guarded numeric addition crates/perry-codegen/src/expr/binary.rs, crates/perry-codegen/src/expr/mod.rs, crates/perry-codegen/src/expr/null_default_numeric_add_tests.rs, test-files/test_issue_8607_null_default_numeric_add.ts	Null-defaulted locals can use guarded numeric addition. Non-numeric and string cases retain dynamic dispatch. Tests cover IR and runtime behavior.
Array-field cache analysis crates/perry-codegen/src/collectors/proven_this.rs, crates/perry-codegen/src/collectors/mod.rs	Eligible looping methods snapshot stable array fields and rewrite later reads. Mutations, deletes, computed writes, super operations, internal calls, and nested closures prevent or limit caching.
Array-cache clone compilation crates/perry-codegen/src/codegen/artifacts.rs, crates/perry-codegen/src/codegen/method.rs	Code generation emits pointer-array-cache clones with dedicated symbols and propagates the clone flag through lowering calls.
Dispatch routing and regression coverage crates/perry-codegen/src/lower_call/property_get/dynamic_dispatch.rs, crates/perry-codegen/src/collectors/proven_this_routing_tests.rs, changelog.d/8607-pipeline-null-default-add.md	Proven receivers prefer array-cache clones. Aliased receivers continue to use exact-shape routing. Tests cover contained receivers and rejected field rebinding cases.

Estimated code review effort: 4 (Complex) | ~45 minutes

Merge Risk: ⚪ Minimal · up to 68c8e

The change is merge-ready after normal checks and review; no actionable merge-blocking risk remains.

Sequence Diagram(s)

sequenceDiagram
  participant Caller
  participant DynamicDispatch
  participant ArrayCacheClone
  participant PShapeClone
  Caller->>DynamicDispatch: invoke instance method
  DynamicDispatch->>ArrayCacheClone: select cached clone for proven receiver
  DynamicDispatch->>PShapeClone: select shape clone for aliased receiver

Loading

sequenceDiagram
  participant ExpressionLowering
  participant NumericGuard
  participant DynamicAddition
  ExpressionLowering->>NumericGuard: detect null-defaulted value plus literal
  NumericGuard->>DynamicAddition: retain fallback for non-numeric values
  DynamicAddition-->>ExpressionLowering: emit guarded numeric control flow

Loading

Suggested reviewers: thehypnoo, jdalton

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 58.82% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 34 functions across 10 files. (1 skipped: 1 unsupported.)	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly identifies the code generation performance improvement for generic registry pipelines.
Description check	✅ Passed	The description covers the required summary, changes, issue, tests, benchmark results, output, and checklist, with a justified full-suite exception.
Linked Issues check	✅ Passed	The changes address issue #8607 by optimizing pipeline code generation, preserving dynamic behavior, adding tests, and reporting improved wall time with unchanged RSS.
Out of Scope Changes check	✅ Passed	The code, tests, parity fixture, and changelog entry are related to the pipeline performance objectives in issue #8607.

✨ 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]

Merging as a validated batch of three, stacked on current main.

check	result
cargo check --workspace --all-targets	exit 0
perry-codegen --lib	1167 passed, 1 ignored
perry-runtime --lib (RUST_TEST_THREADS=1)	2634 passed, 4 ignored
all seven ratchets	0
cargo fmt --all -- --check	0

Ratchets re-run against the current baseline immediately before merge.

#8610's ratchet debt is resolved. It was held twice for +2 bare raw-handle reads in class_registry/construct.rs (7 against a ceiling of 5) and +1 open-coded StringHeader payload offset. On head c9b994d0f both gates now report clean — 925/925 raw-handle reads, 108 modules within ceilings, payload inventory exit 0. Converted rather than baseline-refreshed, which was the right call given the counts had gone up. Thanks for taking that rather than reaching for --write-baseline; construct.rs runs during instance construction, so a raw handle held across an allocating call there is exactly the shape that yields a stale pointer surfacing cycles later somewhere unrelated.

#8613 and #8614 are aimed at the two tickets filed after the last sweep — #8606 (typed_array round 2, where the wall ratio now exceeds the instruction ratio) and #8607 (pipeline, still 1.74x instructions at ~1.05x cycles). Worth being explicit about what this validation does and does not establish: the gates and suites say these are safe to land, not that they are faster. Both are perf changes, and the corpus sweep after merging is the actual verdict. I will run it and report against those tickets.

Two metadata fixes applied while staging (fork PRs, so they could not be pushed to the branches): #8610 shipped without a changelog.d fragment, so I wrote one; #8614's was keyed 8607- after its issue and is now PR-keyed.