Ask the contract whether the number branch admits an integer; refuse multiples a double cannot spell (0.15.2)

[rasuvaeff]

Fixes #132
Fixes #133

• oneOf over integer and number: the number branch is judged by the old float multipleOf rule, and the contract disagrees since 0.12.1 #132 CompositionArbitraries::numberBranchAdmits() judged the number branch's multipleOf by a copy of the float rule the contract left in 0.12.1; the copy disagreed on 4670 of the integers in ±100000 for 0.7 (58254 kept on the integer branch, seen on both by the contract). The verdict is SchemaCheck::isMultipleOf() (openapi-contract 0.12.2, required ^0.12.2). Zoo amounts.create gains step (0.7 beside a wide integer branch); corpus re-recorded in openapi-contract#156.
• A decimal multiple past the precision of a double is rounded to a non-multiple instead of being refused at compile time #133 a number bound so wide that |bound| × 10^decimals > 2^53 fails closed at compile time; a multiple that divides one (1, 0.5, 0.25) is exempt. A property (300 runs, nine multiples, bounds to 1e9) pins that every generated multiple is one to the contract.
• CI static-analysis.yml runs composer rector on every PR — 0.15.0 shipped rector-red.

composer build   PASS
composer rector  PASS
package-audit    0 errors, 0 warnings

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: d32081c8-cecf-4c92-8dd5-f36ce6407fe2

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}