I'm Bonk, and I've done a quick review of your PR.

This PR reverts the requestTermination / isTerminationRequested mechanism from the C++ JSG layer, but the Rust FFI layer still depends on those removed methods.

Issues (by severity)

1. Build break (high): The Rust FFI layer in src/rust/jsg/ has active callers of the removed IsolateBase::requestTermination() and IsolateBase::isTerminationRequested() methods. These files need corresponding changes:

   • src/rust/jsg/ffi.c++:1089-1095 — isolate_request_termination() and isolate_is_termination_requested() call the removed methods
   • src/rust/jsg/ffi.h:343-344 — declares those FFI functions
   • src/rust/jsg/v8.rs:555-556 — Rust-side FFI declarations
   • src/rust/jsg/lib.rs:599-623 — Lock::request_termination() and Lock::is_termination_requested() public methods
   • src/rust/jsg/lib.rs:819 — catch_panic() calls lock.request_termination() after converting a Rust panic to a JS exception
   • src/rust/jsg-test/tests/resource_callback.rs:441-488 — test that asserts on is_termination_requested()

   The catch_panic case is particularly important: it's the Rust panic handler that terminates the isolate after a panic to prevent GC from observing partially-mutated state. This needs a replacement — likely switching to terminateExecution() (which is still available on IsolateBase) instead of the removed requestTermination(). The is_termination_requested() check in the test would need a different assertion strategy since there's no longer a flag to query.

I'm Bonk, and I've done a quick review of your PR.

This PR reverts the requestTermination/isTerminationRequested mechanism from C++ and Rust, replacing the Rust-side usage with direct terminateExecution() calls, since the upstream V8 fix makes the workaround unnecessary.

The build break I flagged in the previous review has been addressed — the Rust FFI layer now consistently uses isolate_terminate_execution across all three bridge files, catch_panic still terminates the isolate after a Rust panic, and the test assertions on the removed flag have been cleaned up.

Issues

1. (Low) Stale comments in resource_callback.rs — the doc comment on line 441 and the code comment on line 462 still reference request_termination() instead of terminate_execution(). These are outside the diff so I can't suggest inline, but they should be updated for consistency.

This review was generated by an AI assistant and may contain errors.