fix(fetch): cancel the body when toFetchResponse throws

[dinwwwh]

toFetchResponse now cancels the response body when building the Response throws. Before, an event-stream body was already pulling from its iterator when toFetchHeaders or new Response failed, so the iterator was never released (a generator's finally never ran) and the 15s keep-alive interval kept firing for the life of the process.

Fixes

• Bodies are released when the response can't be built: a 204/205/304 status with a body, a status outside 200–599, or an invalid header name/value.
• The original error is still thrown unchanged. A stream that refuses to cancel (locked) no longer surfaces as an unhandled rejection.
• The success path has no extra cost.

Testing

• New tests: an event-stream body with a 204 status, a stream body with an invalid header, and a locked stream. The first two fail against main.
• Full vitest run (1306 passed), plus eslint and tsc -b for packages/fetch, are clean.

🤖 Generated with Claude Code

[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@128

@standard-server/core

npm i https://pkg.pr.new/@standard-server/core@128

@standard-server/fastify

npm i https://pkg.pr.new/@standard-server/fastify@128

@standard-server/fetch

npm i https://pkg.pr.new/@standard-server/fetch@128

@standard-server/node

npm i https://pkg.pr.new/@standard-server/node@128

@standard-server/peer

npm i https://pkg.pr.new/@standard-server/peer@128

@standard-server/shared

npm i https://pkg.pr.new/@standard-server/shared@128

commit: 52bd2e2

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[codspeed]

Merging this PR will not alter performance

✅ 26 untouched benchmarks
⏩ 108 skipped benchmarks¹

---

_{Comparing claude/tofetchresponse-stream-cleanup-3fbd11 (52bd2e2) with main (28db1dd)}

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. ↩

[pullfrog]

✅ No new issues found.

Reviewed changes — reviewed the complete diff of 52bd2e2 (initial review).

• packages/fetch/src/response.ts: wraps new Response(...) and void response.body in a try; on throw, cancels a ReadableStream body via body.cancel(error).catch(() => {}) and rethrows the original error. Success path is byte-for-byte equivalent.
• packages/fetch/src/response.test.ts: three new cases — event-stream body with a 204 status (asserts iterator cleanup runs), stream body with an invalid header value (asserts cancel gets the TypeError), and a locked stream (asserts the original /locked/ error still propagates without an unhandled rejection).

I independently confirmed the tests are discriminating: restoring response.ts to base 28db1dd fails the two cleanup cases, and dropping .catch(() => {}) makes the locked-stream case surface an unhandled ERR_INVALID_STATE. Full vitest run (1306 passed), tsc -b for packages/fetch, and eslint on both changed files are clean. The event stream is the only body source that starts pulling before new Response, and toFetchBody cannot throw after toEventStream returns, so the leak is fully covered by the new try.

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