Repository navigation
feat: land shared rules, modern desktop UI and offline Windows baseline - #55
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
@codex review |
|
Codex Review: Didn't find any major issues. 🚀 Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
@codex review |
|
Codex Review: Didn't find any major issues. 🎉 Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
@codex review |
💡 Codex Reviewhttps://github.com/HemSoft/yahtzee/blob/d9074e02cbd9ec0bbe4dec2121d7ed8b19c3aae7/packages/game-engine/src/scoring.ts#L227-L231 When this is deployed over existing Convex data, verified ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
Addressed the persisted-score finding on e5fd65d. Additive version fields and indexes isolate historical rankings/history and prevent new top-ten eviction from deleting old-rule results. Old active sessions reject without writes rather than silently mixing rules. Backend regressions, full quality measurement, refreshed local Convex bindings and portable qualification pass. No production migration or deployment was run. Current-head screenshots and recording were recaptured and replaced in Validation. |
|
@codex review |
|
Codex Review: Didn't find any major issues. 👍 Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
e5fd65d to
5babbc5
Compare
|
@codex review |
|
Codex Review: Didn't find any major issues. Keep it up! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Summary
Land the previously requested desktop/web redesign, shared scoring updates, and portable offline Windows edition. This establishes the shared-engine baseline needed by #40 while preserving the existing online clients. It does not implement or release the iOS app.
Acceptance criteria
Definition of done
This is the inherited baseline integration from #40, not completion of that issue. Public iOS identity, native configuration, durable mobile resume, device qualification, and App Store approval remain open in #38 and its children. No signing, deployment, release upload or legal declaration is part of this PR.
Validation
Head
5babbc506717a19dffcaf5871b85a5065142e2f2, Windows, Node 24.12.0 and declared Bun 1.4.2 throughnpx --package bun@1.4.2.bun run typecheck,bun run lint,bun run testandbun run securitypass. 184 tests, 976 assertions.bun run quality:measurepasses, including existing web, Electron and mobile-web-adapter journeys plus desktop/responsive design assertions. These are not native iOS tests.bun run mutation:proberejects the weakened assertion at the unchanged 95% gate. Full mutation qualification passes at 97.5207%. Its command now includes the new suggestion/offline tests; the initial omission produced a real 91.74% failure that was fixed rather than waived.bun run package:offlineandbun run test:offlinepass. The actual portable EXE launches from a path with spaces, completes six/eight/ten-dice games with networking disabled, restarts with scores retained, and supports play-again and quit.tests/clients/design.spec.tsand the portable EXE smoke test. A recording captures the shared responsive design journey, not a native iOS device.Setup, light appearance, 1440 by 1000:
Compact desktop, 944 by 685, six dice and two players. Grand total remains visible:
Responsive web, dark appearance, 390-pixel viewport. Held states and full scorecard remain reachable:
Twenty dice and four players. This expanded scorecard intentionally scrolls:
Portable EXE setup and saved results:
video.webm
Hosted Windows initially exposed a real 655-pixel usable-height overflow. The new regression fails before the fix and passes afterward. Both setup, including recent names, and active gameplay now have explicit 934-by-655 regressions. Compact panel/header and setup-form spacing now fits without shrinking score buttons or hiding overflow. The complete client, packaging, portable EXE and recording checks were repeated on this head.
Risks and deferred work
The Windows EXE is unsigned and may trigger SmartScreen. Windows closing/quitting abandons an active game; only completed results persist. Native mobile resume is separate work in #41. The existing game name is not cleared for public distribution; #39 remains a release blocker. The MIT license does not grant third-party trademark rights. No claim of App Store readiness is made.
Scoring-version compatibility
Verified historical totals stay in their original partition. New sessions/results use rules version 2, and ranking, top-ten eviction and history queries partition before applying limits. Historical results remain accessible with the explicit version-1 selector. The additive schema requires no data rewrite. Old active sessions reject without writes and ask the player to start again. See deployment and rollback guidance. No production data was changed or deployment performed. Regression tests cover separate history/rankings, preserved historical rows, versioned results and pre-upgrade session rejection.