Previously, this behavior was tested indirectly by the REPL tests.
However, REPL now uses DONT_CONTEXTIFY for its VM context.

PR-URL: #62571
Refs: #62371
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>

Previously, this behavior was tested indirectly by the REPL tests.
However, REPL now uses DONT_CONTEXTIFY for its VM context.

PR-URL: #62571
Refs: #62371
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>

Previously, this behavior was tested indirectly by the REPL tests.
However, REPL now uses DONT_CONTEXTIFY for its VM context.

PR-URL: nodejs#62571
Refs: nodejs#62371
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>
Reviewed-By: Michaël Zasso <targos@protonmail.com>

test-vm-global-contextual-store was added upstream in nodejs/node#62571. It
asserts that an undeclared strict-mode contextual store (`"use strict"; z = 42`)
throws a ReferenceError. Chromium's V8 removed
v8::PropertyCallbackInfo<T>::This(), so Electron's contextify setter cannot
distinguish a contextual store from an explicit `globalThis.x = v` store; it
keeps the original behavior (writing to the sandbox) so explicit global stores
in vm modules keep working (test-vm-module-referrer-realm). Guard the
strict-mode ReferenceError assertion in the new test under Electron.

Ref: nodejs/node#62571
Ref: nodejs/node#60616
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

test-vm-global-contextual-store was added upstream in nodejs/node#62571. It
asserts that an undeclared strict-mode contextual store (`"use strict"; z = 42`)
throws a ReferenceError. Chromium's V8 removed
v8::PropertyCallbackInfo<T>::This(), so Electron's contextify setter cannot
distinguish a contextual store from an explicit `globalThis.x = v` store; it
keeps the original behavior (writing to the sandbox) so explicit global stores
in vm modules keep working (test-vm-module-referrer-realm). Guard the
strict-mode ReferenceError assertion in the new test under Electron.

Ref: nodejs/node#62571
Ref: nodejs/node#60616
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

This new upstream test asserts that a strict-mode store to an undeclared
global in a vm context throws ReferenceError. Chromium's V8 contextify
global-proxy interceptor instead creates the property, so the assertion
is fundamentally incompatible with Electron's V8 and cannot be guarded
to pass.

Ref: nodejs/node#62571

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* chore: bump node in DEPS to v24.16.0

* chore: remove upstreamed patch

Node.js restored fs patchability in the ESM loader upstream, making
the fix_lazyload_fs_in_esm_loaders_to_apply_asar_patches.patch
obsolete (the patch's exact change is now in lib/internal/modules/
esm/{load,resolve,translators}.js).

Ref: nodejs/node#62835

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 72638db)

* fix(patch): re-add experimental_fetch member after upstream cleanup

Upstream removed the experimental_fetch field from EnvironmentOptions,
but Electron's patch still registers --experimental-fetch as a CLI
option bound to that field. Re-add the member so the option compiles.

Ref: nodejs/node#62759

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 6fd9370)

* fix(patch): cast const away when freeing uv_cpu_info_t.model

libuv 1.52.1 typed uv_cpu_info_t.model as const char*, but uv__free
takes void*. Electron builds with -Werror,-Wincompatible-pointer-types-
discards-qualifiers, so add a cast. The memory is allocated via strdup
so the cast is safe.

Ref: nodejs/node#61829

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 90d1009)

* fix(patch): silence sign-compare warning in sessionVarintGetSafe
Cast int nBuf to size_t when comparing with sizeof(aCopy) so the
bundled sqlite3 amalgamation compiles under -Werror,-Wsign-compare.

Ref: nodejs/node#62699

Co-Authored-By: Claude <noreply@anthropic.com>

* test: move root package.json aside in node spec runner

third_party/electron_node lives under Chromium's src/, whose package.json
("type": "module") is always an ancestor of the Node.js test tree. Upstream
assumes no package.json sits above the tests, so that ancestor changes how
test files and fixtures resolve their module type: it disables module-syntax
detection (breaking test-compile-cache-typescript-esm) and emits
MODULE_TYPELESS_PACKAGE_JSON warnings that break tests asserting clean stderr
(test-esm-detect-ambiguous, test-esm-import-meta-main-eval,
test-output-coverage-with-mock).

Move src/package.json aside for the duration of the run so the environment
matches upstream exactly, then restore it. The original is kept in a sibling
backup file so an interrupted/killed run self-heals on the next invocation
rather than leaving src/package.json missing.

Ref: Unable to locate reference

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 29abd0c)

* fix(patch): mark test-macos-app-sandbox as flaky

The test copies the Electron binary into a standalone .app bundle and
code-signs it; under parallel suite runs this races with dyld resolving
the Electron Framework rpath and intermittently aborts (SIGABRT). It
passes reliably when run alone. Mark it flaky so flakes don't fail CI.

Ref: Unable to locate reference

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 3853529)

* test: disable test-tls-set-default-ca-certificates-extra-override

setDefaultCACertificates() round-trips the default CA set through
BoringSSL's X509_STORE, which dedups a duplicate-subject root (DigiCert
Global Root CA) that OpenSSL keeps. The set therefore loses one cert on
re-add (149 -> 148), so the test's assertEqualCerts round-trip check
fails under Electron's BoringSSL. The sibling -recovery test is disabled
for the same reason.

Ref: nodejs/node#58822

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 3dbc596)

* fix(patch): mark test-runner watch tests as flaky

test-run-watch-repeatedly, test-run-watch-run-duration and
test-run-watch-without-file race under parallel suite load: the watcher
fires an extra re-run before the assertion, so the expected single-run
output ("tests 1") arrives with accumulated subtests. All three pass in
isolation.

Ref: nodejs/node#44898

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit 0f9c327)

* chore(patches): update libuv const-cast patch management

Combine the Windows libuv cpu_info const-cast update into the existing
chore_cast_const_away_when_freeing_uv_cpu_info_t_model.patch and keep
a single patch-management commit for the final exported patch series.

Co-Authored-By: Claude <noreply@anthropic.com>
(cherry picked from commit c590614)

* fix(patch): add UV_LOOP_INTERRUPT_ON_IO_CHANGE option

Ref: libuv/libuv#3308

Co-Authored-By: Claude <noreply@anthropic.com>

* chore: update patches (trivial only)

Co-Authored-By: Claude <noreply@anthropic.com>

* chore: update patches

Co-Authored-By: Claude <noreply@anthropic.com>

* fix(patch): adapt GetIsolate removal for the inspector NetworkAgent

Adapts api_remove_deprecated_getisolate.patch for the inspector NetworkAgent:
adapt the GetIsolate removal to the NetworkAgent member helpers, qualify the
v8 Maybe helpers in network_agent, and drop duplicate inspector helper
definitions.

Ref: nodejs/node#62162

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* fix(patch): guard vm contextual-store strict assertion for Chromium's V8

test-vm-global-contextual-store was added upstream in nodejs/node#62571. It
asserts that an undeclared strict-mode contextual store (`"use strict"; z = 42`)
throws a ReferenceError. Chromium's V8 removed
v8::PropertyCallbackInfo<T>::This(), so Electron's contextify setter cannot
distinguish a contextual store from an explicit `globalThis.x = v` store; it
keeps the original behavior (writing to the sandbox) so explicit global stores
in vm modules keep working (test-vm-module-referrer-realm). Guard the
strict-mode ReferenceError assertion in the new test under Electron.

Ref: nodejs/node#62571
Ref: nodejs/node#60616
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: John Kleinschmidt <jkleinsc@electronjs.org>
Co-authored-by: Claude <noreply@anthropic.com>