fix: validate native scoring and share the replay engine with WASM - #2
Merged
Merged
Conversation
…; MSRV 1.88 for let chains
Mechanical clippy --fix of collapsible_if and manual_is_multiple_of, which clippy 1.98 raises once the MSRV allows let chains; no logic change. Checked with clippy 1.98.1 -D warnings, cargo fmt --check and cargo +1.88.0 check --all-targets.
…on the whole-live simulation)
…tests' into feat/chart-stats-gekisou-skills
Co-Authored-By: Claude Code <noreply@anthropic.com>
…racts Add synthetic fixtures for band variants, empty hosts, score/tail moments, seed stopping, mission gates and null outputs. Document single-skill limits and the unchanged baseline. The rare-probability regression intentionally exposes the pending structural-randomness fix. Co-Authored-By: Claude Code <noreply@anthropic.com>
…tests' into feat/chart-stats-gekisou-skills
Co-Authored-By: Claude Code <noreply@anthropic.com>
Cover both library entry points, CLI zero cross seeds, missing plain-kind cross terms and cumulative Just indicators. Clarify structural determinism and sampling limits. Co-Authored-By: Claude Code <noreply@anthropic.com>
…tests' into feat/chart-stats-gekisou-skills
Correct scoring gates, pool timing, rank snapshots and Just-window wiring against original ARM64 captures. Check both aptitude precision endpoints. Expose the same per-note replay API to Rust and WebAssembly, with public validation summaries and strict resource-independent comparison tools. Co-Authored-By: Claude Code <noreply@anthropic.com>
nichinichisou0609
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The whole-live model now preserves the native scoring gates, inactive pool timing, Solo and Network rank snapshots, and Just-window state. The same per-note replay API runs in Rust and WebAssembly; aptitude sampling checks both the highest-judgement and Perfect precision targets.
Validation: 272 release regressions, clippy with warnings denied, formatting, and Rust 1.88 compatibility pass. The documented offline ARM64 comparisons cover mixed judgements, frame rates, real ordinary/Gekisou/support skills, probability branches, life changes and ranking modes. Rust and WASM agree for 340 charts × two modes × two input plans. The public validation table records each comparison's actual scope and resource identities.
The statistics and single-skill increments remain separate from formation search. Native checks run the original IL2CPP instructions with explicit host adapters; they do not claim complete device or server equivalence. No game binaries, original master/chart blobs or private traces are committed.
🤖 Generated with Claude Code