P1: перевести Pair на общий joint graph (генерик joint) - #372
Merged
Conversation
|
Important Review skippedReview was skipped as selected files did not have any reviewable changes. 💤 Files selected but had no reviewable changes (1)
⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: ASSERTIVE Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
You can disable this status message by setting the Use the checkbox below for a quick retry:
WalkthroughPairFill и PairLabel переведены на frozen frontend с единой pointwise joint-цепочкой, WCAG 2.2 constraints и fresh recheck. Удалены старые pair-side и H-K production-пути, обновлены валидация, тесты, golden outputs и WASM budget metadata. ChangesЕдиный Pair frontend
Estimated code review effort: 5 (Critical) | ~120 minutes Sequence Diagram(s)sequenceDiagram
participant ThemeConfig
participant SemanticResolver
participant PairFrontend
participant PointwiseJointPointProgramV1
participant JointPointEvaluatorV1
ThemeConfig->>SemanticResolver: validates PairLabel opaque alpha
SemanticResolver->>PairFrontend: lowers PairFill and PairLabel
PairFrontend->>PointwiseJointPointProgramV1: builds candidates and constraints
PointwiseJointPointProgramV1->>JointPointEvaluatorV1: assesses visible occurrences
JointPointEvaluatorV1-->>PointwiseJointPointProgramV1: returns hard decisions
PointwiseJointPointProgramV1-->>PairFrontend: returns feasible selection
PairFrontend-->>SemanticResolver: returns verified emitted paints
Possibly related PRs
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
This was referenced Jul 21, 2026
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@crates/labcolors-core/src/solve.rs`:
- Line 228: Update the documentation comment in solve.rs around the `bg_luma`
reference to remove the broken intra-doc link, replacing it with plain text or a
valid existing symbol while preserving the surrounding explanation.
In `@docs/decisions/0003-hk-scope.md`:
- Around line 19-20: В разделе решения про Pair после P1 замените англоязычные
термины «fill/label occurrences», «WCAG constraints» и «point graph algebra» на
русские эквиваленты, сохранив исходный смысл утверждения и остальную
формулировку без изменений.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 5361b744-8aba-4272-811f-b5aabf7c9824
📒 Files selected for processing (18)
crates/labcolors-core/src/agnostic_gates.rscrates/labcolors-core/src/appearance.rscrates/labcolors-core/src/config.rscrates/labcolors-core/src/config/tests.rscrates/labcolors-core/src/constraints/mod.rscrates/labcolors-core/src/constraints/wcag22.rscrates/labcolors-core/src/exposure_support.rscrates/labcolors-core/src/joint.rscrates/labcolors-core/src/joint_tests.rscrates/labcolors-core/src/lib.rscrates/labcolors-core/src/lpc.rscrates/labcolors-core/src/pair.rscrates/labcolors-core/src/pair_label_tests.rscrates/labcolors-core/src/semantic.rscrates/labcolors-core/src/solve.rscrates/labcolors-core/tests/data/labui_emission_golden.txtdocs/decisions/0003-hk-scope.mddocs/empirical-inventory.md
💤 Files with no reviewable changes (3)
- crates/labcolors-core/src/appearance.rs
- docs/empirical-inventory.md
- crates/labcolors-core/src/exposure_support.rs
Comment on lines
+19
to
+20
| - Pair после P1 не использует H-K: его fill/label occurrences и WCAG constraints | ||
| исполняются общей point graph algebra. H-K не влияет на скрытый выбор стороны. |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Замените англоязычные описательные термины русскими.
Фрагмент содержит fill/label occurrences, WCAG constraints и point graph algebra, что нарушает правило русской терминологии для docs/**.
Предлагаемая формулировка
-- Pair после P1 не использует H-K: его fill/label occurrences и WCAG constraints
- исполняются общей point graph algebra. H-K не влияет на скрытый выбор стороны.
+- После P1 путь Pair не использует H-K: его вхождения заливки и метки, а также
+ ограничения WCAG исполняются общей алгеброй графа точек. H-K не влияет на
+ скрытый выбор стороны.As per path instructions, документация в docs/** должна быть на русском и проверяться на терминологию без англицизмов.
📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| - Pair после P1 не использует H-K: его fill/label occurrences и WCAG constraints | |
| исполняются общей point graph algebra. H-K не влияет на скрытый выбор стороны. | |
| - После P1 путь Pair не использует H-K: его вхождения заливки и метки, а также | |
| ограничения WCAG исполняются общей алгеброй графа точек. H-K не влияет на | |
| скрытый выбор стороны. |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/decisions/0003-hk-scope.md` around lines 19 - 20, В разделе решения про
Pair после P1 замените англоязычные термины «fill/label occurrences», «WCAG
constraints» и «point graph algebra» на русские эквиваленты, сохранив исходный
смысл утверждения и остальную формулировку без изменений.
Source: Path instructions
Заменяет предыдущую попытку (WCAG-в-joint) корректным по DAG дизайном из #370 (ветка agent/p1-correct-rebuild @ 08b2e82, старый агент). Joint остаётся генерик-solver'ом (Pointwise*<E>, инстанс ExactSrgb8IdentityV1) и НЕ импортирует WCAG-specific payload — по V2a exit-критерию «production joint не импортирует WCAG payload». Readability-floor label остаётся в замороженном semantic/legacy слое до F1/R1, а не зашивается в ядро solver'а. - fill Paint -> fill-on-page occurrence -> surfaceFrom -> label-on-fill occurrence; - PairSide / pair_fill / resolve_pair_label и pair-физика удалены; - public Rust/WASM/TS/FFI surface без изменений; .github и Swift-гейт не тронуты. Локально: fmt --all --check, clippy --all-targets -D warnings, 19 core test-бинарников и workspace build — зелёные. WASM size-budget сброшен на baseline (386547): размер дизайна #370 будет переизмерен на CI и записан отдельным коммитом (raw локально 875313 vs baseline 831652 — на ~6KB меньше прежнего WCAG-в-joint варианта). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LzZyjuo5ahzmW6V8pPJqSJ
lemone112
force-pushed
the
claude/color-engine-semantic-drift-jadrtr
branch
from
July 21, 2026 03:09
a8bf45a to
fa53d51
Compare
lemone112
changed the title
…6 B) Дизайн #370 (генерик joint, WCAG снаружи) измерен на CI run 29797754817: runtime artifact 421876 B (gzip 184975). Это честное изменение current measured contract по C3b: maxRawBytes == rawBytes (ноль headroom), ratchet fail-closed. На 6113 B меньше отменённого WCAG-в-joint варианта (427989): WCAG-evaluator больше не тянется в достижимый joint-путь. Основной рост (+35329 B vs baseline) — сам joint-solver и мономорфизация его типов (замерено ранее twiggy). Self-referential guard синхронизирован: WASM_BUDGET_FILE_SHA256 пересчитан. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LzZyjuo5ahzmW6V8pPJqSJ
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@crates/labcolors-core/src/pair.rs`:
- Around line 88-97: Update VerifiedPairV1::execution so the missing-execution
case is handled through the module’s typed error path instead of unreachable! or
another panic. Propagate that result through callers such as ordinal,
fill_paint, and label_paint, or enforce the invariant structurally by storing
the selected JointExecutionRecordV1 during construction while preserving the
existing valid execution behavior.
In `@crates/labcolors-core/src/semantic.rs`:
- Around line 2687-2781: Preserve the typed unreachable result from
select_label_candidates in lower_pair_label_frontend: map WcagInfeasible to the
appropriate Wcag22 unreachable SolveFailure, while continuing to wrap unexpected
selection failures as InternalInvariant. Keep the existing selection context and
validation behavior unchanged.
In `@docs/whitepaper.md`:
- Around line 111-115: В описании pipeline замените англоязычные термины
hard-report, total order, fresh recheck и physical lowering на русские
эквиваленты, сохранив смысл контракта и остальной порядок этапов без изменений.
- Around line 82-83: Переведите англоязычные контрактные формулировки в строках
таблицы `PairFill` и `PairLabel` на русский, включая термины `frozen frontend`,
`exact authored source`, `opaque` и `finite label candidate domain`, сохранив
исходный смысл и формат таблицы.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: aed3ab8e-f80e-4bfb-b2ff-59234339b475
📒 Files selected for processing (26)
crates/labcolors-core/src/agnostic_gates.rscrates/labcolors-core/src/appearance.rscrates/labcolors-core/src/config.rscrates/labcolors-core/src/config/preset.rscrates/labcolors-core/src/config/tests.rscrates/labcolors-core/src/constraints/exact.rscrates/labcolors-core/src/constraints/mod.rscrates/labcolors-core/src/constraints/wcag22.rscrates/labcolors-core/src/exposure_support.rscrates/labcolors-core/src/joint.rscrates/labcolors-core/src/joint_tests.rscrates/labcolors-core/src/lib.rscrates/labcolors-core/src/lpc.rscrates/labcolors-core/src/pair.rscrates/labcolors-core/src/pair_label_tests.rscrates/labcolors-core/src/scale.rscrates/labcolors-core/src/semantic.rscrates/labcolors-core/src/solve.rscrates/labcolors-core/tests/agnostic_production_surface.rscrates/labcolors-core/tests/data/labui_emission_golden.txtcrates/labcolors-core/tests/empirical_inventory.rsdocs/decisions/0003-hk-scope.mddocs/empirical-inventory.mddocs/whitepaper.mdpackages/colors/bench/wasm.jsonscripts/check-wasm-size-budget.mjs
💤 Files with no reviewable changes (3)
- crates/labcolors-core/src/exposure_support.rs
- crates/labcolors-core/src/appearance.rs
- crates/labcolors-core/src/solve.rs
Comment on lines
+88
to
97
| impl VerifiedPairV1 { | ||
| fn execution(&self) -> &JointExecutionRecordV1 { | ||
| let executions = match &self.evidence { | ||
| PairSelectionEvidenceV1::Unconstrained(evidence) => evidence.fresh_executions(), | ||
| PairSelectionEvidenceV1::Wcag22(evidence) => evidence.fresh_executions(), | ||
| }; | ||
| let (mut lo, mut hi) = (0.0_f64, 0.5_f64); | ||
| for _ in 0..48 { | ||
| let mid = 0.5 * (lo + hi); | ||
| if inside(mid) { | ||
| lo = mid; | ||
| } else { | ||
| hi = mid; | ||
| } | ||
| } | ||
| lo | ||
| } | ||
|
|
||
| /// Насыщенная ячейка: (L, h, доля максимальной хромы) → display-байты. | ||
| fn swatch(l: f64, h_rad: f64, rel_c: f64) -> [f64; 3] { | ||
| let c = rel_c * max_chroma(l, h_rad); | ||
| encode_clamped(oklab_to_srgb_linear([l, c * h_rad.cos(), c * h_rad.sin()])) | ||
| executions.first().unwrap_or_else(|| { | ||
| unreachable!("one selected Pair tuple over one case has one execution") | ||
| }) | ||
| } |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
execution() паникует на "unreachable" состоянии вместо типизированного возврата.
executions.first().unwrap_or_else(|| unreachable!(...)) — хотя инвариант действительно держится (recheck всегда над 1 candidate / 1 case → execution_count == 1), само правило репозитория требует, чтобы unreachable-состояния возвращались типизированно, а не паниковали. Это приватный helper, но заводить панику на «invariant drift» новым кодом — прямое нарушение декларированной политики.
As per coding guidelines, "Новый или изменяемый public path не должен вызывать panic и не должен получать plausible fallback; invalid, unreachable, unsupported и incomplete context должны возвращаться типизированно."
🛡️ Пример типизированного варианта
-impl VerifiedPairV1 {
- fn execution(&self) -> &JointExecutionRecordV1 {
- let executions = match &self.evidence {
- PairSelectionEvidenceV1::Unconstrained(evidence) => evidence.fresh_executions(),
- PairSelectionEvidenceV1::Wcag22(evidence) => evidence.fresh_executions(),
- };
- executions.first().unwrap_or_else(|| {
- unreachable!("one selected Pair tuple over one case has one execution")
- })
- }
+impl VerifiedPairV1 {
+ fn execution(&self) -> Result<&JointExecutionRecordV1, PairInvariantDriftV1> {
+ let executions = match &self.evidence {
+ PairSelectionEvidenceV1::Unconstrained(evidence) => evidence.fresh_executions(),
+ PairSelectionEvidenceV1::Wcag22(evidence) => evidence.fresh_executions(),
+ };
+ executions.first().ok_or(PairInvariantDriftV1)
+ }Далее вызывающие методы (ordinal/fill_paint/label_paint/...) пробрасывают эту ошибку типизированно, либо execution() можно оставить &self без Result, если весь модуль согласится нести инвариант через конструктор (например, хранить JointExecutionRecordV1 напрямую в VerifiedPairV1 при создании, а не искать .first() каждый раз).
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@crates/labcolors-core/src/pair.rs` around lines 88 - 97, Update
VerifiedPairV1::execution so the missing-execution case is handled through the
module’s typed error path instead of unreachable! or another panic. Propagate
that result through callers such as ordinal, fill_paint, and label_paint, or
enforce the invariant structurally by storing the selected
JointExecutionRecordV1 during construction while preserving the existing valid
execution behavior.
Source: Coding guidelines
Comment on lines
+2687
to
2781
| fn lower_pair_label_frontend( | ||
| bg: &BgInput, | ||
| tint: crate::ladder::LadderTint, | ||
| tint: LadderTint, | ||
| fraction: f64, | ||
| floor: Floor, | ||
| surface_alpha_light: f64, | ||
| surface_alpha_dark: f64, | ||
| vc: &ViewingConditions, | ||
| ) -> PendingResolution { | ||
| let alpha = if vc.is_dark_theme() { | ||
| surface_alpha_dark | ||
| } else { | ||
| surface_alpha_light | ||
| }; | ||
| let tint_q = quantise_encoded(tint.for_vc(vc)); | ||
| let surface_hex = | ||
| match crate::alpha::composite_hex_from_encoded(tint_q, alpha, bg.encoded_display()) { | ||
| Ok(hex) => hex, | ||
| Err(error) => { | ||
| return Err(SolveFailure::InvalidInput(format!( | ||
| "тинт-поверхность бейджа вне encoded-sRGB8 reference-домена: {error}" | ||
| ))); | ||
| } | ||
| }; | ||
| let Ok(surface_bg) = BgInput::solid(&surface_hex) else { | ||
| return Err(SolveFailure::InternalInvariant( | ||
| "тинт-поверхность бейджа вне кодированного домена sRGB".into(), | ||
| if surface_alpha_light.to_bits() != 1.0_f64.to_bits() | ||
| || surface_alpha_dark.to_bits() != 1.0_f64.to_bits() | ||
| { | ||
| return Err(SolveFailure::InvalidInput( | ||
| "PairLabel surface representation must be opaque after P1".into(), | ||
| )); | ||
| }; | ||
| } | ||
|
|
||
| let source = tint.srgb8_for_vc(vc); | ||
| let backdrop = pair_root_surface(bg)?; | ||
| let fill = lower_pair_fill_occurrence(bg, tint, vc)?; | ||
| let surface = fill.visible(); | ||
| let surface_bg = BgInput::solid(&surface.to_hex()).map_err(|error| { | ||
| SolveFailure::InternalInvariant(format!("generated PairFill Surface was rejected: {error}")) | ||
| })?; | ||
| let surface_ctx = ResolveContext::new(&surface_bg, vc); | ||
| let anchor = TextAnchor::new(fraction, floor)?; | ||
| resolve_hued_anchor(&surface_bg, anchor, tint, vc, &surface_ctx) | ||
|
|
||
| let mut resolved_candidates = Vec::new(); | ||
| let mut physical_candidates = Vec::new(); | ||
| let mut order = Vec::new(); | ||
|
|
||
| match pair_candidate(&surface_bg, fraction, Floor::None, source, vc, &surface_ctx) { | ||
| Ok((resolved, bytes)) => { | ||
| let ordinal = crate::joint::CandidateOrdinalV1::new(1); | ||
| resolved_candidates.push((ordinal, resolved)); | ||
| physical_candidates.push(crate::pair::PairLabelCandidateV1::new(ordinal, bytes)); | ||
| order.push(ordinal); | ||
| } | ||
| Err(error) if matches!(floor, Floor::None) => return Err(error), | ||
| Err(error) if error.boundary().is_none() => return Err(error), | ||
| Err(_) => {} | ||
| } | ||
|
|
||
| if !matches!(floor, Floor::None) { | ||
| let (resolved, bytes) = | ||
| pair_candidate(&surface_bg, fraction, floor, source, vc, &surface_ctx)?; | ||
| if physical_candidates | ||
| .iter() | ||
| .all(|candidate| candidate.source() != bytes) | ||
| { | ||
| let ordinal = crate::joint::CandidateOrdinalV1::new(2); | ||
| resolved_candidates.push((ordinal, resolved)); | ||
| physical_candidates.push(crate::pair::PairLabelCandidateV1::new(ordinal, bytes)); | ||
| order.push(ordinal); | ||
| } | ||
| } | ||
|
|
||
| if physical_candidates.is_empty() { | ||
| return Err(SolveFailure::InternalInvariant( | ||
| "Pair frontend produced an empty candidate domain".into(), | ||
| )); | ||
| } | ||
|
|
||
| let verified = crate::pair::select_label_candidates( | ||
| source, | ||
| crate::composition::AdmittedOpacityV1::OPAQUE, | ||
| physical_candidates, | ||
| order, | ||
| backdrop, | ||
| pair_requirement(floor), | ||
| ) | ||
| .map_err(|error| { | ||
| SolveFailure::InternalInvariant(format!( | ||
| "Pair joint selection failed after typed proposal admission: {error:?}" | ||
| )) | ||
| })?; | ||
|
|
||
| if verified.fill_occurrence() != fill.occurrence() | ||
| || verified.fill_paint() != fill.paint() | ||
| || verified.label_occurrence().visible() != verified.label_paint().source().bytes() | ||
| { | ||
| return Err(SolveFailure::InternalInvariant( | ||
| "Pair joint evidence drifted from the selected physical chain".into(), | ||
| )); | ||
| } | ||
|
|
||
| resolved_candidates | ||
| .into_iter() | ||
| .find(|(ordinal, _)| *ordinal == verified.ordinal()) | ||
| .map(|(_, resolved)| resolved) | ||
| .ok_or_else(|| { | ||
| SolveFailure::InternalInvariant( | ||
| "Pair selection returned an ordinal outside the proposal domain".into(), | ||
| ) | ||
| }) | ||
| } |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Сравнить формулу контраста в WCAG22-эвалюаторе joint-модуля с legacy wcag.rs
fd -a 'wcag22.rs' crates/labcolors-core/src
fd -a 'joint.rs' crates/labcolors-core/src
rg -n -A15 'struct Wcag22Srgb8V1' crates/labcolors-core/src/wcag22.rs crates/labcolors-core/src/joint.rs 2>/dev/null
rg -n -A10 'impl.*Wcag22Srgb8V1' crates/labcolors-core/src/wcag22.rs crates/labcolors-core/src/joint.rs 2>/dev/null
rg -n -B2 -A10 'fn contrast_ratio' crates/labcolors-core/src/wcag.rsRepository: Labpics-Team/lab-colors
Length of output: 1458
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Map relevant symbols and inspect the exact error path.
ast-grep outline crates/labcolors-core/src/semantic.rs --view expanded >/tmp/sg_semantic.txt || true
ast-grep outline crates/labcolors-core/src/joint.rs --view expanded >/tmp/sg_joint.txt || true
ast-grep outline crates/labcolors-core/src/constraints/wcag22.rs --view expanded >/tmp/sg_wcag22.txt || true
ast-grep outline crates/labcolors-core/src/wcag22.rs --view expanded >/tmp/sg_wcag22_top.txt || true
printf '\n== semantic excerpt ==\n'
sed -n '2700,2775p' crates/labcolors-core/src/semantic.rs
printf '\n== joint.rs relevant symbols ==\n'
rg -n -A6 -B4 'select_label_candidates|PairLabelCandidateV1|CandidateOrdinalV1|boundary\(' crates/labcolors-core/src/joint.rs
printf '\n== constraints/wcag22.rs relevant symbols ==\n'
rg -n -A8 -B4 'Wcag22Srgb8V1|Wcag22CriterionV1|contrast_ratio|infeasible|criterion' crates/labcolors-core/src/constraints/wcag22.rs
printf '\n== wcag22.rs top-level ==\n'
rg -n -A8 -B4 'Wcag22Srgb8V1|Wcag22CriterionV1|contrast_ratio|infeasible|criterion' crates/labcolors-core/src/wcag22.rsRepository: Labpics-Team/lab-colors
Length of output: 21451
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '1,260p' crates/labcolors-core/src/joint.rsRepository: Labpics-Team/lab-colors
Length of output: 8755
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect pair-specific requirements and the selection error surface.
rg -n -A8 -B8 'fn pair_requirement|enum Floor|fn pair_candidate|select_label_candidates\(' crates/labcolors-core/src/semantic.rs crates/labcolors-core/src/pair.rs crates/labcolors-core/src/joint.rsRepository: Labpics-Team/lab-colors
Length of output: 5246
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '1,260p' crates/labcolors-core/src/pair.rsRepository: Labpics-Team/lab-colors
Length of output: 9209
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect the candidate construction and the exact Wcag22 infeasible/error paths.
sed -n '2664,2795p' crates/labcolors-core/src/semantic.rs
printf '\n== pair errors in pair.rs ==\n'
rg -n -A6 -B4 'ExactInfeasible|WcagInfeasible|ExactRecheck|WcagRecheck|Infeasible\(report\)' crates/labcolors-core/src/pair.rsRepository: Labpics-Team/lab-colors
Length of output: 7460
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Find whether SolveFailure has a typed pair infeasible variant or similar paths.
rg -n -A4 -B4 'enum SolveFailure|WcagInfeasible|ExactInfeasible|PairLoweringErrorV1|InternalInvariant' crates/labcolors-core/srcRepository: Labpics-Team/lab-colors
Length of output: 43970
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect the surrounding public API and whether pair-label failures are mapped
# to a typed SolveFailure elsewhere.
rg -n -A5 -B5 'lower_pair_label_frontend|PairLoweringErrorV1|FloorUnreachable|WcagInfeasible|select_label_candidates\(' crates/labcolors-core/srcRepository: Labpics-Team/lab-colors
Length of output: 28189
Не сворачивайте Wcag22-недостижимость в InternalInvariant crates/labcolors-core/src/semantic.rs:2757-2761 — если select_label_candidates возвращает WcagInfeasible для обычного no-solution case, это должно выходить как типизированная unreachable-ошибка, а не как внутренний баг.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@crates/labcolors-core/src/semantic.rs` around lines 2687 - 2781, Preserve the
typed unreachable result from select_label_candidates in
lower_pair_label_frontend: map WcagInfeasible to the appropriate Wcag22
unreachable SolveFailure, while continuing to wrap unexpected selection failures
as InternalInvariant. Keep the existing selection context and validation
behavior unchanged.
Comment on lines
+82
to
+83
| | `PairFill` | frozen frontend: exact authored source как opaque Paint/Occurrence | | ||
| | `PairLabel` | finite label candidate domain против фактически emitted PairFill Surface | |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Переведите контрактные формулировки на русский.
Новые строки содержат frozen frontend, exact authored source, opaque и finite label candidate domain, что нарушает требования к русскоязычной документации и терминологии без англицизмов.
Предлагаемая формулировка
-| `PairFill` | frozen frontend: exact authored source как opaque Paint/Occurrence |
-| `PairLabel` | finite label candidate domain против фактически emitted PairFill Surface |
+| `PairFill` | замороженный фронтенд: точный авторский источник как непрозрачные Paint/Occurrence |
+| `PairLabel` | конечная область кандидатов метки относительно фактически выданной поверхности PairFill |📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| | `PairFill` | frozen frontend: exact authored source как opaque Paint/Occurrence | | |
| | `PairLabel` | finite label candidate domain против фактически emitted PairFill Surface | | |
| | `PairFill` | замороженный фронтенд: точный авторский источник как непрозрачные Paint/Occurrence | | |
| | `PairLabel` | конечная область кандидатов метки относительно фактически выданной поверхности PairFill | |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/whitepaper.md` around lines 82 - 83, Переведите англоязычные контрактные
формулировки в строках таблицы `PairFill` и `PairLabel` на русский, включая
термины `frozen frontend`, `exact authored source`, `opaque` и `finite label
candidate domain`, сохранив исходный смысл и формат таблицы.
Sources: Coding guidelines, Path instructions
Comment on lines
+111
to
+115
| `PairLabel` использует фактически emitted `PairFill` occurrence как derived | ||
| Surface. Frontend формирует конечный label candidate domain; общий joint engine | ||
| исполняет `fill → surfaceFrom → label`, строит полный hard-report, выбирает по | ||
| явному total order и повторяет fresh recheck. Имена клиентских токенов и | ||
| `FillPrimary` в physical lowering не участвуют. |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Уберите англоязычные термины из описания pipeline.
hard-report, total order, fresh recheck и physical lowering оставляют ключевую часть контракта на английском. Используйте русские эквиваленты; например, «явно заданный общий порядок» и «повторная проверка свежими данными».
🧰 Tools
🪛 LanguageTool
[uncategorized] ~113-~113: Слово пишется через дефис: «по-явному».
Context: ...l`, строит полный hard-report, выбирает по явному total order и повторяет fresh recheck. ...
(Pravopisanie_po-prezhnemu)
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@docs/whitepaper.md` around lines 111 - 115, В описании pipeline замените
англоязычные термины hard-report, total order, fresh recheck и physical lowering
на русские эквиваленты, сохранив смысл контракта и остальной порядок этапов без
изменений.
Sources: Coding guidelines, Path instructions
…(30 leaves) P1 убирает старый pair_fill притемняющий nudge: brand/danger/info accent-fill эмитируются как сырые identity-якоря. 30 листьев resolveVars (5 fill-ключей x 6 фонов, light-тема) обновлены на фактический выход движка. recheck-секция байт-идентична (0/2752 drift, замерено локально). Значения сверены собранным pkg через engine.resolveTheme; полный npm-suite зелёный (dev-deps установлены). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LzZyjuo5ahzmW6V8pPJqSJ
Прежняя фраза утверждала «ни один fill не изменён» (верно для glow-регена ADR-0004, но ложно после P1). Дополнена правдой: 30 fill-leaves перешли на identity-якорь, pair_fill nudge удалён, recheck 0/2752 drift. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LzZyjuo5ahzmW6V8pPJqSJ
This was referenced Jul 21, 2026
Collaborator
Author
Canonical execution trackerThe post-P1 hardening program, file-ownership rules and recovery procedure are now centralized in #399. P1 remains intentionally isolated: do not fold #386/#393/#382/#378, #395/#392/#394/#397 or other optimization slices into this PR. After merge, record the merge SHA and newly released file ownership in #399. |
This was referenced Jul 21, 2026
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
2 participants
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.
Что это
Чистый P1 поверх C8c: лоуверит Pair в общий генерик joint graph.
Ключевое исправление относительно #365/#366
jointостаётся генерик-solver'ом (Pointwise*<E>, инстансExactSrgb8IdentityV1) и НЕ импортирует WCAG-specific payload — по V2a exit-критерию «production joint не импортирует WCAG payload». Прежняя попытка зашивалаJointHardConstraintV1::Wcag22прямо в ядро solver'а — это дрейф семантики в core; здесь он убран. Readability-floor label остаётся в замороженномsemantic/legacy слое до F1/R1, а не в generic-ядре.Дизайн взят из #370 (
agent/p1-correct-rebuild@08b2e828, старый агент); здесь доведён до зелёного на моей ветке.Инварианты
PairSide/pair_fill/resolve_pair_labelи pair-физика удалены;lower → surfaceFrom → upper;.githubи Swiftnative-conformancegate целы;Согласовано с DAG
V2a + C8b → P1; предпосылки G1a/V1a/V2a/C8b вmain. WCAG-в-joint убран строго по V2a exit-критерию.Статус
Локально зелёное:
fmt --all --check,clippy --workspace --all-targets -D warnings, 19 core test-бинарников, workspace build (core/wasm/ffi/conformance). WASM size-budget переизмеряется на CI и будет записан отдельным коммитом.🤖 Generated with Claude Code
https://claude.ai/code/session_01LzZyjuo5ahzmW6V8pPJqSJ
Summary by CodeRabbit
Новые возможности
Исправления