DO NOT MERGE: sandbox benches

[NathanFlurry]

Description

Please include a summary of the changes and the related issue. Please also include relevant motivation and context.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Please describe the tests that you ran to verify your changes.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[NathanFlurry]

• chore: add rivetkit-native cross-platform builds to pkg-pr-new #4600
• DO NOT MERGE: sandbox benches #4597 👈 (View in Graphite)
• fix(envoy-client): add tls #4595 : 1 other dependent PR (#4599 )
• chore: fix driver test suite #4591
• chore: tunnel auth #4588
• chore: remove udb as a dependency of envoy-client #4587
• chore: revert actor v2 hack #4586
• fix: fix pool metadata protocol version refresh, rivetkit native ws #4585
• feat: US-001 - Define SqliteKv trait in rivetkit-sqlite-native #4584
• feat(rust/engine-runner): minimal rust engine runner #4289
• chore: delete kv channel #4583
• feat(engine): rust runner #4582
• chore: remove non-engine drivers #4574 : 1 other dependent PR (#4581 )
• feat: dynamic actors #4397
• fix: driver test suite & sqlite reliability #4559
• fix: remove serverless token #4572
• fix: misc token fixes #4569
• fix(pb): clean up actor stop decision handling #4566
• fix(envoy): use global instance, add signal handlers #4565
• chore: misc fixes, add pb snapshot test #4561
• fix(cache): make in memory cache global #4560
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[claude]

Code Review: DO NOT MERGE: sandbox benches

This PR is marked DO NOT MERGE so these are early-stage notes.

Security: /api/bench has no auth (examples/sandbox/src/server.ts:7-10). The endpoint is publicly accessible on the deployed Cloud Run instance. Anyone who knows the URL can trigger benchmark runs against actor infrastructure. Guard with an internal secret or source IP restriction before deploying to a shared environment.

Bug: startServerlessActor is now fire-and-forget (wrapper.js around line 101). The function was changed from async (payload) => await jsHandle.startServerless(...) to a sync void wrapper with no return value. If the caller awaits it expecting a promise, errors will now be silently dropped. Verify callers do not need the returned promise.

Docker: COPY . . copies the full monorepo (examples/sandbox/Dockerfile). build-sandbox.sh correctly builds Docker from .sandbox-deploy/, but the Dockerfile itself has no guard against being used directly on the repo root. Add a comment clarifying the expected build context.

Build: silent WASM fallback (scripts/docker/build-sandbox.sh:74-78). When the .node binary is missing, the script prints a warning and continues, producing an image that falls back to WASM silently. Since this PR benchmarks native SQLite, a silent fallback means measuring the wrong backend. Consider exit 1 here so CI catches it.

Duplicated callAction: implemented identically in bench.ts, bench-sqlite.ts, bench-wake.ts, and bench-endpoint.ts. Fine for throwaway scripts, but any gateway URL or header change requires four edits.

Minor: package.json is missing a trailing newline (diff shows "No newline at end of file").

Minor: Cargo.toml adds rdkafka as a workspace dependency but no package in this PR uses it. Add it to the consuming package directly or include the consumer in this PR.

Observation: isPlainObject and toNativeBinding were deleted from wrapper.js. Confirm these are unused outside wrapper.js and that binding conversion is now fully handled on the Rust NAPI side.

---

Overall the benchmarking structure is solid. Separating client-side scripts from server-side bench-endpoint.ts is clean, and subtracting baseline RTT to isolate actor processing time from network variance is the right approach.

[claude]

PR Review: sandbox benches. This PR is marked DO NOT MERGE and is an experimental benchmarking branch. CRITICAL: Hardcoded Credentials in bench-throughput.ts - staging credentials are hardcoded in plaintext, must be replaced with env vars. CODE QUALITY: wrapper.js respondCallback calls changed to fire-and-forget causing silent error swallowing. Debug console.log statements left in websocket_open path. startServerlessActor made fire-and-forget losing startup errors. ARCHITECTURE: Benchmark definitions duplicated across bench.ts/bench-remote.ts/bench-endpoint.ts. Dockerfile assumes pre-built bundle from build-sandbox.sh. engine-api-full pinned to temporary pkg.pr.new tarball not workspace. zod v3 to v4 migration in lock file may break transitive deps. SUMMARY: Good benchmarking coverage but hardcoded credentials and fire-and-forget lifecycle changes must be fixed before merge.