test262 regression on main — parity 96.5%→95.8%, 322 tests (bisecting)

[proggeramlug]

test262 regression on main — parity 96.5% → 95.8%, 322 tests newly failing

A full test262 sweep of main at the #6008 tip (76cc2b996) regressed vs the #5979 tip (28948ea44):

metric	#5979 (good)	#6008 (bad)
parity	96.5%	95.8%
Temporal self-validated	99.1%	96.7%
pass	35242	34981
runtime-fail	1164	1421

322 tests newly failing, 61 newly fixed (net −261). Harness-confirmed (not a measurement artifact): e.g. built-ins/Boolean/prototype/toString dropped 100% → 67%.

Newly-failing by area

• built-ins/Temporal 159
• language/expressions 119
• language/arguments-object 15, built-ins/Proxy 5, Boolean 3, Number 3, Function 2, TypedArray 2, others 14

Reason clusters

• TypeError: method is not a function — 131
• (no output) (crash/SIGSEGV) — 126
• Expected SameValue(x, y) — 35
• Cannot read properties of undefined — 10

Regression window

28948ea44 (good) .. 76cc2b996 (bad) — 21 commits. Broad-impact suspects: #6004 (native_call_method.rs method-dispatcher change), #5978/#5981 (GC), #5988 (async-step deadlock class). Bisect in progress.

Sample newly-failing cases

built-ins/Boolean/prototype/toString/S15.6.4.2_A2_T1.js
built-ins/Number/prototype/toString/S15.7.4.2_A4_T01.js
built-ins/Function/prototype/S15.3.5.2_A1_T2.js
built-ins/Temporal/Duration/compare/order-of-operations.js
built-ins/Temporal/Duration/from/order-of-operations.js
built-ins/Proxy/getOwnPropertyDescriptor/trap-is-undefined-target-is-proxy.js
annexB/language/global-code/block-decl-global-block-scoping.js

Measured on an internal Linux sweep host (built-ins + language + intl402 + annexB, jobs 6). A bisect+fix is underway; this issue tracks the root-cause + fix/revert PR.

[proggeramlug]

Bisected. Culprit = #5874 "Improve Zod compatibility support" (849f297a8).

git bisect (good 28948ea44 .. bad 76cc2b996), witness slice = built-ins/Boolean/prototype/toString + Number/prototype/toString + Temporal/Duration/{compare,from}:

commit	witness parity	Boolean	Temporal.Duration	self-validated
4f5bc6adb (#5981)	98.3%	89%	100%	100%
849f297a8 (#5874) — first bad	93.9%	67%	95%	95.1%
76cc2b996 (main tip)	93.9%	67%	95%	95.1%

(532853f4a between them is a metadata-only chore.)

#5874 is a 156-file monolithic squash that swept in unreviewed codegen/runtime WIP + two 5.1 MB binaries (constv, direct) + ~30 stray tests/*.sh alongside the intended Zod work. Two swept changes dominate:

• codegen/lower_call/property_get/number_string.rs gated the universal .toString() interception on is_numeric_expr || is_bigint_expr, so boxed built-in receivers (new String()/new Number()/new Date()) with a user-assigned built-in method bypassed the brand-check and stopped throwing TypeError → the "method is not a function" / SameValue clusters. (property_get.rs reclassified trim/case-conversion the same way.)
• The runtime prototype-identity / class-registry / instanceof rewrite regressed Temporal subclass prototype resolution (Object.getPrototypeOf(result) === construct.prototype) → the Temporal 159 + (no output) crash clusters.

Fix = #6015 (inverse-patch of the #5874 payload rebased onto current main; #5991 preserved; #5983 stdlib change untouched). Witness slice returns to 98.3% / self-validated 100% / Temporal.Duration 100%. Supersedes the stale #5994.

[proggeramlug]

Resolved by #6015 — regression fully healed

A full re-sweep of main at the #6015 tip confirms the 322-test regression is reverted with the grind wins preserved:

metric	pre-regression	regressed (#6008)	now (#6015)
parity	96.5%	95.8%	96.7%
Temporal self-validated	99.1%	96.7%	99.1%
pass	35242	34981	35288
runtime-fail	1164	1421	1114

Culprit was #5874's contaminated squash (unreviewed codegen/runtime WIP swept in alongside the Zod work); #6015 inverse-patched it out while preserving the legitimate later commits (#5991/#5999/#5983). Temporal self-validation and the Boolean/Number/Function brand-check negatives are all restored; net parity is now slightly above the pre-regression baseline. Closing.