hardening and polish sweep

cheapnud commented

2026-08-13 15:14:11 +00:00

Owner

Closes the pricing engine audit stack and the module polish backlog: code validation, resolve guards, rounding coverage, a default-address backstop index, and assorted small fixes across purchasing, planning, addresses, export and materials.

cheapnud added 6 commits

2026-08-13 15:14:11 +00:00

validate variant code characters and pin serialization order d17d0996b2

guard fabric resolution and cover subtotal rounding fecc700cbb

polish purchasing and planning edge cases 13fd862f70

polish addresses export and materials edges with a default backstop index c1185babbb

make the default address swap atomic under the backstop index bbbd80ef4d

surface publish warnings on the price versions page

Claude PR Review / AI Code Review (pull_request) Successful in 17s

Details

Build & Test / Build and Test (pull_request) Successful in 7m6s

Details

55beb8168a

forgejo-actions commented

2026-08-13 15:14:28 +00:00

🤖 Claude AI Code Review

Summary
Well-scoped hardening sweep adding validation guards, warning surfacing, a filtered unique index backstop, and extensive test coverage across pricing, purchasing, planning, and addresses.
Code Quality ⭐⭐⭐⭐⭐
Exceptionally thorough — every change is backed by tests and reasoned comments, and the SetDefaultDeliveryAddressAsync ordering fix shows real care against the new index.
Issues

🟢 MaterialResolution.cs — Enum.IsDefined(priceGroup.Kind) uses the object overload, which boxes on every call. Minor allocation in a pricing hot path; use the generic Enum.IsDefined<FabricMaterialKind>(...) if targeting .NET 5+.

🟢 PartyService.SetDefaultDeliveryAddressAsync — the two-save-within-a-transaction workaround relies on EF change-tracker flush order being "load order," which is not a documented contract. It works and is now transactional, but a cleaner approach would be a single ExecuteUpdate to clear siblings. Not blocking.

Verdict: APPROVE

Solid, test-backed sweep with no correctness or safety concerns. The migration correctly reconfigures the existing index, and the address-default ordering bug fix is a genuine catch.

📊 Tokens: 81382 input + 473 output | 💰 Cost: ~$0.4187
Generated by Claude Opus 4.8 via Gitea Actions

## 🤖 Claude AI Code Review 1. **Summary** Well-scoped hardening sweep adding validation guards, warning surfacing, a filtered unique index backstop, and extensive test coverage across pricing, purchasing, planning, and addresses. 2. **Code Quality** ⭐⭐⭐⭐⭐ Exceptionally thorough — every change is backed by tests and reasoned comments, and the `SetDefaultDeliveryAddressAsync` ordering fix shows real care against the new index. 3. **Issues** 🟢 `MaterialResolution.cs` — `Enum.IsDefined(priceGroup.Kind)` uses the object overload, which boxes on every call. Minor allocation in a pricing hot path; use the generic `Enum.IsDefined<FabricMaterialKind>(...)` if targeting .NET 5+. 🟢 `PartyService.SetDefaultDeliveryAddressAsync` — the two-save-within-a-transaction workaround relies on EF change-tracker flush order being "load order," which is not a documented contract. It works and is now transactional, but a cleaner approach would be a single `ExecuteUpdate` to clear siblings. Not blocking. 4. **Verdict**: **APPROVE** Solid, test-backed sweep with no correctness or safety concerns. The migration correctly reconfigures the existing index, and the address-default ordering bug fix is a genuine catch. --- 📊 **Tokens**: 81382 input + 473 output | 💰 **Cost**: ~$0.4187 *Generated by Claude Opus 4.8 via Gitea Actions*

cheapnud merged commit c80f5e1a4b into master

2026-08-13 15:21:45 +00:00

cheapnud referenced this pull request from a commit

2026-08-13 15:21:46 +00:00

Merge pull request 'hardening and polish sweep' (#4) from d/hd1-hardening into master

Rows
Columns

hardening and polish sweep #4

🤖 Claude AI Code Review