[dinwwwh]

Requested by dinwwwh · project thread

What

This PR adds one bullet to the downgrader README's tool notes, after the kin-openapi bullet.

Some validators reject a number with a zero fraction, such as 1.0, for type: "integer" when they read a 3.0 document:

• Python's openapi-schema-validator (OAS30Validator), which openapi-core uses
• jsonschema-rs with draft 4 rules

That follows how 3.0.3 defined an integer: "a JSON number without a fraction or exponent part". 3.0.4, the version the output names, counts 1.0 as an integer ("both 1 and 1.0 are equivalent"). 3.1 does too, and so do the 3.1 rules of both tools.

Checked

• openapi-schema-validator 0.9: OAS30Validator({type: integer}) rejects 1.0 and 1e0, while OAS31Validator accepts both.
• jsonschema-rs 0.60.3: Draft4Validator rejects 1.0 and Draft202012Validator accepts it.
• JavaScript validators such as Ajv can't tell 1.0 from 1, so they aren't affected.
• The output keeps type: "integer", which code generators need, so the downgrader can't loosen this without turning integers into numbers. That's why this is a docs note only.

The other downgrader thread found this through keyword fuzzing.

---

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

✅ 29 untouched benchmarks
⏩ 2 skipped benchmarks¹

---

_{Comparing claude/project-thread-ttohne (2e60837) with main (79f69e4)}

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 PR adds one bullet to the tool notes under the #### Limitations list of the 3.1 → 3.0 schema section in packages/downgrader/README.md.

• Draft 4 integer-fraction note — documents that Python's openapi-schema-validator (OAS30Validator, which openapi-core uses) and jsonschema-rs under draft 4 rules reject a zero-fraction number such as 1.0 for type: "integer", matching how 3.0.3 defined an integer, while 3.0.4 (the version the output names) and 3.1 count it as one, as do both tools' 3.1 rules.

I independently checked the load-bearing facts: OAS 3.0.3 §4.4 ("a JSON number without a fraction or exponent part") and 3.0.4 §4.4 ("both 1 and 1.0 are equivalent, and are both considered to be integers"); jsonschema 4.10.3 (Draft4Validator rejects 1.0, Draft202012Validator accepts); jsonschema-rs's draft-4 is_integer (uses is_written_as_integer, with tests asserting 1.0/1e0/1e2 are not integers under draft 4); and out.openapi = '3.0.4' in v3.1-to-v3.0.ts. The note is accurate and reads consistently with the neighboring tool bullets.

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