[proggeramlug]

Summary

Part of the intl402/NumberFormat worklist (#5904). Fixes the largest coherent single-root subcluster: built-in Intl functions must not have an own prototype property and must not be constructable.

The Intl.NumberFormat method/getter/bound functions — the format accessor, the bound [[BoundFormat]] function, resolvedOptions, formatToParts, formatRange, formatRangeToParts, and the static supportedLocalesOf — were created via js_closure_alloc without being flagged non-constructable. That let function_would_have_own_prototype lazily synthesize a .prototype object for them and left them dynamically new-able, unlike ordinary built-ins (Math.max, Array.prototype.map) which correctly report hasOwnProperty("prototype") === false and isConstructor === false.

Per ECMA-262 §17, a built-in function only gets the auto-created .prototype when it implements [[Construct]]. The fix marks the closures created by the three shared Intl installation helpers as non-constructable — the same flag the native-module builtin methods already carry:

• install_function — prototype methods (formatToParts/resolvedOptions/…), the static supportedLocalesOf, and the this-based instance methods (formatRange/formatRangeToParts).
• install_bound_instance_function — the bound instance methods (format/formatToParts/resolvedOptions).
• the accessor-getter path in install_constructor — the get …prototype.format accessor.

The constructor closures themselves are untouched, so new Intl.NumberFormat() and Reflect.construct(Intl.X, args, Custom) still work.

Tests fixed (test262, intl402/NumberFormat)

prototype/format/builtin.js
prototype/format/format-function-builtin.js
prototype/formatRange/builtin.js
prototype/formatRangeToParts/builtin.js
prototype/resolvedOptions/builtin.js
supportedLocalesOf/builtin.js

Verification

Same-methodology A/B on the pinned test262 slice (PERRY_NO_AUTO_OPTIMIZE=1, one perry binary, base commit differing only by this edit):

• intl402/NumberFormat: 217 → 223 pass, 31 → 25 runtime-fail. The 6 fixed cases are exactly the builtin.js cluster above.

• Full intl402 slice: 3100 → 3120 pass, 228 → 208 runtime-fail, 0 → 0 diff, 2 → 2 compile-fail — 20 cases fixed, zero regressions. Because the fix lives in the shared installation helpers, the same class of assertion now passes across every Intl service:

  NumberFormat   6  (format builtin, format-function-builtin, formatRange,
                     formatRangeToParts, resolvedOptions, supportedLocalesOf)
  DateTimeFormat 6  (format, format-function, formatRange, formatRangeToParts,
                     resolvedOptions, supportedLocalesOf)
  Collator       3  (compare-function-builtin, resolvedOptions, supportedLocalesOf)
  PluralRules    2  (resolvedOptions, supportedLocalesOf)
  DurationFormat 2  (format not-a-constructor, formatToParts not-a-constructor)
  Intl           1  (supportedValuesOf builtin)
  

Added a per-PR-visible gap test (test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts) that asserts the full ECMA-262 §17 surface ([object Function] tag, extensibility, [[Prototype]] === Function.prototype, no own prototype, not a constructor) for the NumberFormat functions plus DateTimeFormat.prototype.resolvedOptions and Collator.supportedLocalesOf; byte-identical to node --experimental-strip-types.

Code-only PR (no version bump / CHANGELOG / CLAUDE.md edit) per the worklist instructions.

Summary by CodeRabbit

• Bug Fixes

  • Corrected Intl built-in functions so they no longer expose an own prototype property when they are not constructable.
  • Aligned constructability behavior across Intl formatters, prototype methods, accessors, and locale utilities.
  • Preserved normal functionality for formatting, resolved options, and locale-related operations.

• Tests

  • Added regression coverage for Intl function descriptors, constructability, and runtime behavior.

[coderabbitai]

📝 Walkthrough

Walkthrough

Intl builtin installation now marks non-constructor functions as non-constructable, preventing own prototype properties. A regression test checks this behavior across bound methods, accessors, static methods, related Intl services, and constructors.

Changes

Intl builtin constructability

Layer / File(s)	Summary
Mark Intl builtin closures crates/perry-runtime/src/intl.rs, crates/perry-runtime/src/intl/install.rs	Bound methods, shared builtin functions, and Intl prototype accessor getters are marked non-constructable.
Verify Intl builtin behavior test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts	Tests confirm non-constructor Intl functions lack own prototype properties while remaining callable, and constructors retain constructability.

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

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly states the core fix: Intl built-in functions no longer expose an own prototype.
Description check	✅ Passed	The description is detailed and covers the fix, affected Intl APIs, tests, and verification, even though some template sections are omitted.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.

✨ 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.}

[coderabbitai]

🧹 Nitpick comments (1)

test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts (1)

1-6: 🎯 Functional Correctness | 🔵 Trivial

Run the gap suite with Node 26.5.0. The pinned runtime in .node-version is 26.5.0, and the current Node is v24.15.0.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts` around lines 1 - 6,
Run the gap suite for test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts
using Node 26.5.0, matching the version pinned in .node-version; do not use the
current Node v24.15.0.

Source: Coding guidelines

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts`:
- Around line 1-6: Run the gap suite for
test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts using Node 26.5.0,
matching the version pinned in .node-version; do not use the current Node
v24.15.0.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 5241fa30-5829-4da0-8e65-2cfcccc7139c

📥 Commits

Reviewing files that changed from the base of the PR and between 3fd379f and 8e9d031.

📒 Files selected for processing (3)

• crates/perry-runtime/src/intl.rs
• crates/perry-runtime/src/intl/install.rs
• test-files/test_gap_intl_builtin_fn_no_prototype_5904.ts

[proggeramlug]

Thanks @coderabbitai. On the Node-version nitpick: the gap test only exercises ECMA-262 §17 built-in-function invariants ([object Function] tag, extensibility, [[Prototype]] === Function.prototype, absence of an own prototype, isConstructor) plus two locale-stable en-US outputs (1,234.5, en-US). None of these are Node-version-sensitive — they're identical across 24.x/26.x. I validated it byte-for-byte against local Node 26.3.0, and conformance-smoke runs it against the pinned 26.5.0 (.node-version), which is the authoritative gate here. No code change needed.

[coderabbitai]

Tip

For best results, initiate chat on the files or code changes.

@proggeramlug Thanks for the detailed clarification. Agreed—the assertions are specification-level invariants plus locale-stable en-US results, and the pinned Node 26.5.0 conformance-smoke gate is authoritative. The Node-version concern does not apply here; no change is needed.

🐇 ✓

---

✏️ Learnings added

Learnt from: proggeramlug
URL: https://github.com/PerryTS/perry/pull/6707

Timestamp: 2026-07-20T06:53:58.022Z
Learning: For the Intl built-in-function regression gap test, assertions cover ECMA-262 §17 function invariants and locale-stable `en-US` outputs only. Its relevant conformance gate is `conformance-smoke`, which runs under the Node version pinned in `.node-version` (Node 26.5.0 for PR `#6707`); do not flag separate Node 24.x/26.x compatibility concerns for this test without evidence of version-sensitive behavior.

Note: Learnings are effective only in the context of similar code segments. To apply general coding guidelines, please configure review instructions. You can manage existing learnings in the app.