[dinwwwh]

Requested by dinwwwh · project thread

Summary

In 3.1, when an Encoding Object sets style, explode, or allowReserved, its contentType (explicit or default) is ignored, and that holds for multipart/form-data too. 3.0 ignores those three in multipart and applies contentType instead. So a multipart part that 3.1 accepts in any type was held to one type after the downgrade, which makes the 3.0 document stricter.

The fix is a finish hook on Encoding Objects: when any of the three is set, contentType becomes */*, which accepts any part type. Everything else passes through as before. The README gains one table row, and the limitation about 3.0 ignoring style/explode/allowReserved in multipart is narrowed to the one gap left (the untyped part's application/octet-stream default).

Tradeoffs

• A code generator that reads contentType to pick a part's serialization sees */* instead of, say, application/json. Clients that relied on it will send the part as raw bytes or text. That loses a hint, but no value 3.1 accepts is rejected. oRPC does not write style/explode/allowReserved in encodings, so its output is unchanged.

Test plan

• New test covering style, explode, and allowReserved each, and an Encoding Object without them left as is.
• vitest run packages/downgrader (300 passed) and pnpm check pass locally.

---

Generated by Claude Code

[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

1 warning

⚠️ 1 benchmark is dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

✅ 29 untouched benchmarks
⏩ 2 skipped benchmarks¹

---

_{Comparing claude/project-thread-h3q238 (98c8199) with main (b441d6a)}

Footnotes

1. 2 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 — this run reviewed the 3.1→3.0 Encoding Object change in 98c8199, verified against the OAS 3.1.0 and 3.0.3 Encoding Object text.

• finishEncoding on Encoding Objects — MEDIA_TYPE_FIELDS.encoding now routes each encoding through convertObject(..., ENCODING_FIELDS, finishEncoding), which sets out.contentType = '*/*' iff the source explicitly sets style, explode, or allowReserved. This matches 3.1's "value of contentType (implicit or explicit) SHALL be ignored" rule for all three fields, and 3.0's handling of them as ignored for non-x-www-form-urlencoded media types, so */* is the loosest correct representation of the unconstrained part.
• Test coverage — the new test exercises explode, style, and allowReserved: false (each beside or without an explicit contentType) plus a control part with none of the three, which passes through unchanged.
• README — one table row is added, and the "Encoding Objects pass through" limitation is narrowed to the one remaining gap (an untyped multipart part's application/octet-stream default).

The finishEncoding presence test (!== undefined) correctly treats false as defined, plain assignment is the established idiom for constant literal keys in this file, and 3.2→3.1 needs no mirror fix since 3.2 carries the same contentType-ignored rule.

  ^{｜ View workflow run ｜ Using deepseek-v4.1-flash (free via Pullfrog for OSS) ｜ 𝕏}