[dinwwwh]

The aws-lambda sendStandardResponse now releases the response body when setting up the response stream throws synchronously. Before, awslambda.HttpResponseStream.from() or the prelude-triggering res.write('') could throw. The promise rejected, but the body toNodeHttpBody had already built was never destroyed. An event-stream generator never ran its finally and its keep-alive interval kept firing; a ReadableStream body was never cancelled. On a warm Lambda container these leaks carried over into later invocations.

The setup now runs inside a try, like the node adapter. The catch destroys the body with destroyNodeHttpBody, destroys the response stream instead of leaving it half-open, and rejects with the original error.

Behaviour

• Setup errors now clean up the body. Examples: the runtime's Cannot set content-type, too late. when the stream was already written to (canWriteToNodeResponse can't detect that on the Lambda stream), a missing awslambda global, or a throwing first write.
• The error/close listeners that settle the promise are now attached to responseStream before the try. The Lambda runtime's from() returns the same stream it is given, so this is the same object as before. It also means a throwing from() can't leave the error event from destroy unhandled.
• The success path is unchanged: the stream branch keeps its own close listener that destroys the body.
• Removed the WARNING: errors that occur here are silently ignored comment that was copied from the node adapter. It isn't true here: res.destroy(error) emits error on the response stream, which rejects the promise, as the existing body-error test checks.

Testing

• New it.each covering three cases: from throws, the first write throws, and awslambda is missing. Each asserts the rejection, responseStream.errored, and that the body was destroyed. All three fail against main.
• vitest run packages/aws-lambda (100 passed), eslint packages/aws-lambda, and tsc -b are clean.

🤖 Generated with Claude Code

https://claude.ai/code/session_01AVoginFKKdE3QrCKUARJ98

[pkg-pr-new]

• @standard-server/hono-node-playground
• @standard-server/node-http-playground

@standard-server/aws-lambda

npm i https://pkg.pr.new/@standard-server/aws-lambda@126

@standard-server/core

npm i https://pkg.pr.new/@standard-server/core@126

@standard-server/fastify

npm i https://pkg.pr.new/@standard-server/fastify@126

@standard-server/fetch

npm i https://pkg.pr.new/@standard-server/fetch@126

@standard-server/node

npm i https://pkg.pr.new/@standard-server/node@126

@standard-server/peer

npm i https://pkg.pr.new/@standard-server/peer@126

@standard-server/shared

npm i https://pkg.pr.new/@standard-server/shared@126

commit: fa4baf5

[codspeed]

Merging this PR will not alter performance

✅ 26 untouched benchmarks
⏩ 108 skipped benchmarks¹

---

_{Comparing claude/peaceful-heisenberg-88d4hw (fa4baf5) with main (ddd0a09)}

Footnotes

1. 108 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[pullfrog]

✅ No new issues found.

Reviewed changes

• Response setup wrapped in try/catch: sendStandardResponse now catches synchronous failures from toLambdaHeaders, HttpResponseStream.from(), the prelude res.write(''), and end(), tears the body down via destroyNodeHttpBody, destroys the response stream, and rejects with the thrown error.
• Listeners attached up front: the error/close handlers moved onto responseStream ahead of setup, so a throwing from() still has an error listener, and the single close handler now performs the guarded body teardown for every path instead of only the stream branch.
• New failure tests: three it.each cases (from throws, first write throws, awslambda global missing) assert rejection, responseStream.errored === error, and a destroyed body.

I also confirmed the new cases are discriminating: reverting response.ts to b5b4035 while keeping the new tests makes all three fail with responseStream.errored still null, so they genuinely pin the cleanup. Once the tree is restored, the package tests, type:check, and lint are all clean.

  ^{｜ View workflow run ｜ Using DeepSeek Flash (default — pick a model for stronger reviews) ｜ 𝕏}

[pullfrog]

✅ No new issues found.

Reviewed changes

The only change since the prior pullfrog review (ed8585d) is one behavior-preserving refactor (fa4baf5):

• Body teardown back in the stream-branch close listener: responseStream.once('close', …) now only resolves, while the guarded resBody.destroy(getNodeResponseError(res) ?? undefined) moved into the object-body else branch. This restores the success path to its pre-PR shape and mirrors the node adapter's structure.
• Setup-error cleanup unchanged: the try/catch still calls destroyNodeHttpBody(resBody, error, reject), destroys the response stream, and rejects — so the fix the PR exists for is intact.

Since the runtime's from() returns the same stream it is handed, res === responseStream, making the relocation behavior-equivalent for every path. pnpm vitest run packages/aws-lambda/src/response.test.ts passes 13/13.

  ^{｜ View workflow run ｜ Using DeepSeek Flash (default — pick a model for stronger reviews) ｜ 𝕏}