fix(runtime): OrdinarySet rejects a receiver-owned accessor instead of firing its setter

[proggeramlug]

Summary

OrdinarySetWithOwnDescriptor (sec-ordinarysetwithowndescriptor) step 2 handles the case where the descriptor found during the OrdinarySet walk is a data property (or absent → CreateDataProperty) and the write must land on a Receiver that differs from the object walked. In that tail the Receiver's own descriptor decides the outcome:

Receiver own descriptor	result
accessor (2.d.i)	return false — setter NOT called
non-writable data (2.d.ii)	return false
writable data (2.d.iii)	redefine { value }
none (2.e)	CreateDataProperty

Perry's create_or_update_receiver_property tail invoked the receiver's setter for the accessor sub-case (2.d.i) and returned true. So a Reflect.set(target, k, v, receiverWithSetter) where target[k] is a data property mis-fired the receiver's setter and reported success. The receiver's setter must only run when it is the descriptor found by the OrdinarySet walk itself — served by the separate Accessor arm in ordinary_set_with_receiver, never through this CreateDataProperty-on-receiver tail. The accessor sub-case now returns false without calling the setter.

var target = { x: 1 };
var setterCalled = 0;
var receiver = { set x(v) { setterCalled++; } };
Reflect.set(target, "x", 2, receiver);  // before: true / setterCalled 1 ; node & now: false / 0

Validation

Built and tested on an internal Linux sweep host (--release).

• test262 built-ins/Reflect + Proxy + Object (3875 cases): byte-identical pass/fail between baseline and patched — 0 regressions.
• built-ins/TypedArray + TypedArrayConstructors (2184 cases): 0 regressions; combined with the typed-array [[Set]] fix merged in fix(runtime): typed-array integer-indexed [[Set]] spec rejection + coercion #6059 this flips Set/key-is-valid-index-reflect-set.js to passing (a typed-array element source with an accessor-bearing receiver — the exact data-source / accessor-receiver shape above). On its own the fix touches no other test262 verdict in these dirs but is a prerequisite for that case and is spec-correct.

cargo fmt --all -- --check clean; address-classification audit passes; proxy.rs stays under 2000 lines.

Summary by CodeRabbit

• Bug Fixes
  • Adjusted property update behavior for receiver-owned accessor properties so they are no longer treated as assignable in this path.
  • Prevents attempts to write through an accessor setter when the receiver already owns that property, returning a failure result instead.

[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: b3b7a9c9-3777-4c12-ad3f-0835ff795be0

📥 Commits

Reviewing files that changed from the base of the PR and between ca0e895 and 404ab65.

📒 Files selected for processing (1)

• crates/perry-runtime/src/proxy.rs

---

📝 Walkthrough

Walkthrough

In create_or_update_receiver_property, the OwnSetDescriptor::Accessor match arm no longer invokes call_setter_with_receiver; it now returns false immediately, treating a receiver-owned accessor descriptor as non-settable via this path.

Changes

Proxy Set Path Fix

Layer / File(s)	Summary
Receiver accessor descriptor handling crates/perry-runtime/src/proxy.rs	The OwnSetDescriptor::Accessor branch in create_or_update_receiver_property no longer calls the setter and returns false immediately instead.

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

Possibly related PRs

• PerryTS/perry#5164: Modifies the same OwnSetDescriptor::Accessor handling in create_or_update_receiver_property to reject setter-less accessor paths.
• PerryTS/perry#5882: Adjusts related Proxy set dispatch and receiver propagation impacted by this same accessor-handling behavior.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main behavioral fix in OrdinarySet for receiver-owned accessors.
Description check	✅ Passed	The description clearly explains the bug, the spec correction, and validation results, 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 docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/t262-mech-ordinaryset-accessor

---

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.}