jeremydaly / lambda-api / #218

83%

push

github

Add benchmark suite comparing lambda-api to other Lambda frameworks (#34) (#327)

* Add benchmark suite comparing lambda-api to other Lambda frameworks (#34)

In-process micro-benchmarks that invoke each framework's aws-lambda handler
in the same Node VM with identical synthetic API Gateway (v1 + v2) events,
measuring framework overhead with mitata. A correctness gate validates every
cell before timing.

Frameworks: baseline (raw handler), lambda-api, @vendia/serverless-express,
@fastify/aws-lambda, hono. Scenarios: get-json, path-param, post-json,
routing-50, not-found.

Lives in an isolated benchmarks/ package so the root library keeps its
zero-dependency policy; excluded from the npm tarball by the files whitelist.

A manual/release GitHub workflow refreshes the README Benchmarks section
(marker-delimited, idempotent, with a per-release history row) on each
published release.

* Drop misused @author/@license banner from benchmark file headers

The @author field held the project description instead of an author name,
which read oddly in every file. The descriptive block comment is the useful
part; the boilerplate banner added nothing to these new files.

* Add Middy to the benchmark comparison; drop redundant issue #34 references

- Add a middy framework (@middy/core + http-router + http-json-body-parser +
  http-error-handler) to the comparison set and README history. Middy is a
  middleware engine, so it's wired with the official router/body-parser/error
  handler and otherwise kept a thin layer; documented in the fairness notes.
- Bump the benchmarks Node floor to >=20 (Middy 6 requires it; matches the
  workflow and Fastify 5).
- Remove the redundant 'issue #34' mentions from the benchmark README, run.js,
  and package.json description.

764 of 780 branches covered (97.95%)

Branch coverage included in aggregate %.

673 of 689 relevant lines covered (97.68%)

156.73 hits per line