Summary
Follow-up from the Fastify native performance work: verify whether the shared dispatcher optimization benefits other platform-* adapters, and fix the local platform-bun benchmark route mismatch that currently blocks measurement.
Observed
platform-fastify has a confirmed benchmark improvement from the dedicated adapter optimization plus shared dispatcher change.
- Local
fluo+Bun HTTP comparison returned 404 No handler registered for GET /tenants/t-001/users, so Bun could not be included in the final benchmark gate.
platform-bun and platform-cloudflare-workers use the runtime/web request factory, so the shared dispatcher change may help, but this is not yet measured.
- Other Node-style platforms may have smaller or no impact depending on whether headers are already eagerly snapshotted.
Scope
- Reproduce and fix the
platform-bun benchmark route registration/matching issue.
- Add focused regression coverage for the route behavior.
- Run short targeted benchmark checks for available
platform-* adapters after Bun is measurable.
- Keep changes in a dedicated
.worktrees/ branch.
Out of scope
- Publishing packages locally.
- Merging PRs.
- Broad benchmark harness rewrites unless required for a reliable targeted measurement.
Summary
Follow-up from the Fastify native performance work: verify whether the shared dispatcher optimization benefits other
platform-*adapters, and fix the localplatform-bunbenchmark route mismatch that currently blocks measurement.Observed
platform-fastifyhas a confirmed benchmark improvement from the dedicated adapter optimization plus shared dispatcher change.fluo+BunHTTP comparison returned404 No handler registered for GET /tenants/t-001/users, so Bun could not be included in the final benchmark gate.platform-bunandplatform-cloudflare-workersuse theruntime/webrequest factory, so the shared dispatcher change may help, but this is not yet measured.Scope
platform-bunbenchmark route registration/matching issue.platform-*adapters after Bun is measurable..worktrees/branch.Out of scope