fix: invoke JS subclass property setters

Author: DjDeveloperr

NativeScript/ffi/shared/bridge/HostObjects.mm

@@ -1884,15 +1884,9 @@ NativeApiHostSetResult set(Runtime& runtime, const PropNameID& name, const Value
         class_conformsToProtocol(object_getClass(object_),
                                  @protocol(NativeApiClassBuilderProtocol));
     if (isEngineExtendedInstance) {
-#ifdef NATIVESCRIPT_NATIVE_API_HOST_EXPLICIT_OVERRIDE
       if (invokeEnginePrototypeSetter(runtime, property, value)) {
         NATIVE_API_SET_RETURN(true);
       }
-#else
-      if (enginePrototypeHasSetter(runtime, property)) {
-        NATIVE_API_SET_RETURN(false);
-      }
-#endif
     }
 
     if (Class appearanceClass =

PROGRESS.md

@@ -17,6 +17,44 @@ TypeScript/UI worklets.
 
 ## Latest Update - 2026-06-30
 
+### 2026-06-30 02:42 EDT - JS subclass setters invoke prototype accessors directly on V8
+
+- Scope check:
+  - Still generic NativeScript runtime behavior required by the RN-module
+    branch. No RNS-specific native implementation and no direct-engine backend
+    refactor work.
+  - Simulator-only rule remains active; no physical devices were used.
+- CI finding:
+  - GitHub Actions run `28424196438` on `241c9943` kept setup, FFI boundary
+    check, native build, CLI build, and macOS tests green.
+  - iOS still had one failure:
+    `ApiTests.js SpecialCaseProperty_When_CustomSelector_ImplementedInJS`.
+  - `ApiTests.js NSMutableArrayMethods` still passed.
+  - The failed assertion still showed repeated `getter` output with no
+    `setter:true`, which means V8 did not continue from a native host setter
+    miss into the JS prototype setter for this native-instance path.
+- Change:
+  - JS-extended object property writes now invoke JS prototype setters directly
+    before native metadata/runtime setters on every backend, matching the
+    getter-side direct resolution behavior.
+  - Plain unknown JS fields still use the existing expando/fallback behavior;
+    only an actual JS prototype setter is invoked early.
+  - Source coverage now guards that setter invocation is not gated by
+    `NATIVESCRIPT_NATIVE_API_HOST_EXPLICIT_OVERRIDE` and still happens before
+    native writable-property dispatch.
+- Verification:
+  - Focused construction/property guard test passed:
+    `node packages/react-native/test/runtime-instance-selector-base-dispatch.test.js`.
+  - Runtime RN JS tests passed:
+    `for f in packages/react-native/test/*.test.js; do node "$f" || exit 1; done`.
+  - `git diff --check` passed.
+  - `npm run check:ffi-boundaries` passed.
+  - Existing generated macOS project compile/link passed with
+    `xcodebuild -project dist/intermediates/macos/NativeScript.xcodeproj ...`.
+- Still next:
+  - Commit/push this direct-setter fix and watch fresh PR CI. If iOS goes
+    green, resume the RNS simulator parity sweep from this handoff.
+
 ### 2026-06-30 02:10 EDT - JS subclass setters prefer prototype accessors before native setters
 
 - Scope check:

packages/react-native/test/runtime-instance-selector-base-dispatch.test.js

@@ -104,6 +104,10 @@ for (const relativePath of [
     "enginePrototypeHasSetter(runtime, property)",
     setIndex,
   );
+  const setterGuardIndex = source.lastIndexOf(
+    "#ifdef NATIVESCRIPT_NATIVE_API_HOST_EXPLICIT_OVERRIDE",
+    explicitSetterIndex,
+  );
 
   assert(
     setIndex !== -1 &&
@@ -114,9 +118,9 @@ for (const relativePath of [
       fallbackSetterIndex !== -1 &&
       setIndex < explicitSetterIndex &&
       explicitSetterIndex < writableIndex &&
-      setIndex < fallbackSetterIndex &&
-      fallbackSetterIndex < writableIndex,
-    `${relativePath}: JS-subclassed property writes should prefer prototype setters before native runtime setters on every backend`,
+      setterGuardIndex < setIndex &&
+      writableIndex < fallbackSetterIndex,
+    `${relativePath}: JS-subclassed property writes should invoke prototype setters before native runtime setters on every backend`,
   );
 }