openaienmodel: gpt-5-mini-2025-08-07
v4.12.29 — bug fixes and runtime behavior fixes
Key Points
- merge function headers with per-request headers
- lambda-edge: callback resolution and base64-encoded responses
- aws-lambda: non-identity encodings treated as binary
Summary
This patch release fixes several runtime and edge-case bugs across lambda-edge, AWS Lambda, request header handling, routing, ETag logic, and typings. It also includes small test and docs improvements and contributions from new maintainers. Upgrade if you rely on lambda-edge/AWS-Lambda runtimes or precise header/ETag behavior.
Key Points
- Fix: merge function-level headers with per-request headers so request-specific headers are preserved and combined correctly.
- Fix (lambda-edge): resolve the handler using the callback value and Base64-encode response bodies that were content-encoded, addressing broken responses when running on Lambda@Edge.
- Fix (aws-lambda): treat any non-identity Content-Encoding as binary so encoded responses are handled correctly by AWS Lambda integrations.
- Fix (trie-router): allow an empty wildcard remainder to match after a regexp parameter, fixing certain route edge cases.
- Fix (etag): treat
If-None-Match: *as a match for resources, improving caching semantics. - Fix (types): strip extra properties from array element types in JSONParsed typings to better reflect parsed JSON shapes.
- Chore/docs/tests: minor test script and TypeScript test fixes, add compatibilityDate for workerd tests, and a JSDoc example for the language detector.
- New contributors: first-time contributions included in this release.
Recommended action
- Upgrade to v4.12.29 if you use Lambda@Edge or AWS Lambda integrations, rely on strict ETag semantics, or encountered header-merging/routing issues. Run your integration tests focused on binary/encoded responses and routing edge cases.
Reference
- Full changelog: v4.12.28...v4.12.29
- Release published: 2026-07-10T09:33:30.000Z