fix(validation): preserve representable recovery metrics - #488
fix(validation): preserve representable recovery metrics#488seonghobae wants to merge 828 commits into
Conversation
|
Important Draft PR not reviewedDraft PRs are not automatically reviewed by default.
To automatically review draft PRs, update your CodeRabbit configuration: reviews:
auto_review:
drafts: trueThanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
seonghobae
changed the title
Add crate-level scientific contracts for the power-scale bias-SE rounding, nonnegative RMSE-summary support, and zero-multiplier exact-recovery regressions exposed by exact-head line/branch coverage compilation. No numerical implementation, fixture, expected bit pattern, skip/xfail, or coverage denominator changes. Signed-off-by: Seongho Bae <me@seonghobae.me>
Audit the exact-bias integration crates that participate in line/branch coverage and add crate-level scientific contracts to thirteen remaining files. The docs record represented-input pair-distance, single-rounding, permutation/sign, subtraction-roundoff, and fail-closed underflow invariants without changing fixtures, expected bits, numerical code, skips, or coverage denominators. This prevents coverage compilation from serially discovering the same missing-doc root cause one crate at a time. Signed-off-by: Seongho Bae <me@seonghobae.me>
Refresh the checked-in operator register after the #488 root-cause rustdoc audit. Preserve protected-main and owner boundaries, distinguish hosted RED from proactive source-audit repair, and keep current exact-head queued checks from being promoted to GREEN. Signed-off-by: Seongho Bae <me@seonghobae.me>
Add scientific crate-level contracts for the two integration tests identified by exact-head stable and nightly coverage compilation: RMSE percentile support and overflow-safe bias moments. Preserve behavior, fixtures, expected results, and coverage denominators. Signed-off-by: Seongho Bae <me@seonghobae.me>
Add scientific crate-level contracts for the stable coverage scale-underflow acceptance regression and the nightly Monte Carlo percentile moment-support regression observed on Rust Foundation 34056254247. Preserve fixtures, expected decisions, production arithmetic, coverage denominators, and fail-closed behavior. The scale-underflow file also carries the exact Rust 1.98.0 formatting material from same-head Validation Rustfmt Evidence run 34056251889. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair exact-head Rust Foundation coverage REDs by documenting ValidationReport projection invariants and overflow/subnormal RMSE evidence contracts. Preserve numerical implementation, fixtures, expected decisions, coverage denominators, and fail-closed behavior. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair the next exact-head stable/nightly coverage REDs by documenting nonzero correction collisions, zero-RMSE uncertainty coherence, and the strictly positive all-covered Wilson lower bound. Preserve numerical implementation, fixtures, expected decisions, coverage denominators, and fail-closed projection behavior. The equal-nonzero-correction contract also carries its exact Rust 1.98.0 formatting material from b9792 current-head evidence. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair exact-head stable/nightly coverage REDs by documenting two-replication percentile exhaustion and large-z exact-count Wilson rounding. Preserve numerical implementation, fixtures, expected bits, coverage denominators, and fail-closed evidence semantics. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair current exact-head stable coverage -D missing-docs RED without changing fixtures, numerical behavior, or fail-closed report semantics. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair current exact-head nightly -D missing-docs RED, apply its exact Rust 1.98.0 formatter material, and state that n>16 source-copy arithmetic is characterization rather than independent admission evidence. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair exact-head stable coverage -D missing-docs RED and carry the same-head Rust 1.98.0 formatter material for this file without changing the estimator or fixtures. Signed-off-by: Seongho Bae <me@seonghobae.me>
Repair exact-head nightly -D missing-docs RED without changing the v1 schema, fixtures, projection checks, or fail-closed deserialization semantics. Signed-off-by: Seongho Bae <me@seonghobae.me>
seonghobae
added
bug
ChatGPT Codex Connector
|
@coderabbitai review exact |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
echo '== Exact commit identity =='
git rev-parse HEAD
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
echo '== Focused changed paths =='
git diff --name-status "$parent" "$head"
echo '== Focused diff =='
git diff --find-renames --unified=120 "$parent" "$head" -- \
crates/validation_core/tests/bias_se_resource_isolation.rs \
.github/workflows/bias-se-proof-budget.yml \
crates/validation_core/src/bias_se.rs \
Cargo.toml
echo '== Resource-isolation test outline and contents =='
ast-grep outline crates/validation_core/tests/bias_se_resource_isolation.rs \
--items all --type function,struct,enum,impl 2>/dev/null || true
cat -n crates/validation_core/tests/bias_se_resource_isolation.rs
echo '== Relevant test symbols and process invocations =='
rg -n -C 20 --type rust \
'\b(candidate|fallback|child|Command|process|2047|odd|dyadic|exact|to_bits|bias_standard_error|wide|rss|sample)\b' \
crates/validation_core/tests/bias_se_resource_isolation.rs crates/validation_core
echo '== Exact workflow =='
cat -n .github/workflows/bias-se-proof-budget.yml
echo '== Workflow security, provenance, measurement, and labels =='
rg -n -C 10 \
'permissions:|contents: read|actions/checkout@|actions/upload-artifact@|persist-credentials|EVIDENCE_SHA|pull_request\.head\.sha|github\.sha|git rev-parse|25|sample|isolation|candidate|fallback|RSS|rss|allocator|time -v|artifact' \
.github/workflows/bias-se-proof-budget.yml
echo '== Harness definitions and workflow target references =='
rg -n -C 16 --type rust \
'\b(bias_se_exact_proof_budget|bias_se_resource_isolation|hybrid_narrow_wide_pair|buffered_quadratic|two_pass_quadratic)\b' \
crates Cargo.toml
echo '== Production admission boundaries =='
rg -n -C 18 --type rust \
'\b(pub )?fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error' \
crates/validation_core/src/bias_se.rs crates/validation_core/src/bias.rs
echo '== Exact-head workflow runs =='
gh api "repos/${repo}/actions/runs?head_sha=${head}&per_page=100" \
--jq '.workflow_runs[] | {id,name,event,status,conclusion,head_sha,head_branch,path,created_at,updated_at}'
echo '== Exact-head check runs =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}'
echo '== PR state =='
gh pr view 488 --repo "$repo" --json headRefOid,isDraft,state,mergeStateStatus,statusCheckRollup \
--jq '{headRefOid,isDraft,state,mergeStateStatus,checks:[.statusCheckRollup[]|{name,status,conclusion,detailsUrl}]}'Length of output: 3571 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
parent='6b00d719566b074d2c0b14066efb65eaa8fa296b'
example='crates/validation_core/examples/bias_se_resource_isolation.rs'
workflow='.github/workflows/bias-se-proof-budget.yml'
echo '== Exact example diff and full source =='
git diff --find-renames --unified=160 "$parent" "$head" -- "$example"
ast-grep outline "$example" --items all --type function,struct,enum,impl 2>/dev/null || true
cat -n "$example"
echo '== Workflow source =='
cat -n "$workflow"
echo '== Example and workflow introduction history =='
git log --reverse --format='%H %P %s' -- "$example" "$workflow"
echo '== All changes to the focused files through exact head =='
git log --reverse --format='%H %s' -- "$example" "$workflow" | while read -r sha rest; do
echo "----- $sha $rest -----"
git show --format= --find-renames --unified=80 "$sha" -- "$example" "$workflow"
done
echo '== Exact process, geometry, identity, and measurement references =='
rg -n -C 14 --type rust \
'\b(Command|args|--child|candidate|fallback|2047|odd|dyadic|to_bits|assert_eq|bias_standard_error|hybrid_narrow_wide_pair|buffered_quadratic|two_pass_quadratic)\b' \
"$example" crates/validation_core/examples crates/validation_core/src
echo '== Exact workflow evidence controls =='
rg -n -C 12 \
'permissions:|contents: read|actions/checkout@|actions/upload-artifact@|persist-credentials|EVIDENCE_SHA|pull_request\.head\.sha|github\.sha|git rev-parse|25|sample|isolation|candidate|fallback|RSS|rss|allocator|time -v|artifact' \
"$workflow"
echo '== Production admission source =='
rg -n -C 18 --type rust \
'\bpub fn bias_standard_error\b|match\s+.*len|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error' \
crates/validation_core/src/bias_se.rs crates/validation_core/src/bias.rs
echo '== Exact-head Actions workflow runs =='
gh api "repos/${repo}/actions/runs?head_sha=${head}&per_page=100" \
--jq '.workflow_runs[] | {id,name,event,status,conclusion,head_sha,head_branch,path,run_attempt,created_at,updated_at}'
echo '== Jobs and artifacts for proof-budget runs on the exact head =='
for id in $(gh api "repos/${repo}/actions/runs?head_sha=${head}&per_page=100" --jq '.workflow_runs[] | select(.path == ".github/workflows/bias-se-proof-budget.yml") | .id'); do
echo "----- run $id -----"
gh api "repos/${repo}/actions/runs/${id}/jobs?per_page=100" \
--jq '.jobs[] | {id,name,status,conclusion,head_sha,started_at,completed_at,steps:[.steps[]|{number,name,status,conclusion}]}'
gh api "repos/${repo}/actions/runs/${id}/artifacts?per_page=100" \
--jq '.artifacts[] | {id,name,size_in_bytes,expired,created_at,updated_at,expires_at,digest,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha,head_branch:.workflow_run.head_branch}}'
done
echo '== Exact-head check runs and PR state =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}'
gh pr view 488 --repo "$repo" --json headRefOid,isDraft,state,mergeStateStatus,statusCheckRollup \
--jq '{headRefOid,isDraft,state,mergeStateStatus,checks:[.statusCheckRollup[]|{name,status,conclusion,detailsUrl}]}'Length of output: 203775 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
example='crates/validation_core/examples/bias_se_resource_isolation.rs'
artifact='10163932916'
echo '== Complete exact resource-isolation example =='
git show "${head}:${example}" | nl -ba
echo '== Exact example change =='
git show --format=fuller --find-renames --unified=120 "$head" -- "$example"
echo '== Focused structural calls from the example =='
ast-grep run --lang rust \
--pattern 'Command::new($$$)' "$example" 2>/dev/null || true
ast-grep run --lang rust \
--pattern 'bias_standard_error($$$)' "$example" 2>/dev/null || true
echo '== Immutable artifact file list and contents =='
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
gh api --header 'Accept: application/vnd.github+json' \
"repos/${repo}/actions/artifacts/${artifact}/zip" > "${tmpdir}/artifact.zip"
unzip -Z1 "${tmpdir}/artifact.zip" | sort
unzip -p "${tmpdir}/artifact.zip" \
'candidate.csv' 'fallback.csv' 'isolated-rss-summary.csv' \
'process-resource.txt' 'candidate-process-resource.txt' 'fallback-process-resource.txt' \
'environment.txt' 'SHA256SUMS'
echo '== Artifact digest recheck =='
sha256sum "${tmpdir}/artifact.zip"
gh api "repos/${repo}/actions/artifacts/${artifact}" \
--jq '{id,name,digest,expired,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha}}'Length of output: 31244
I found no finding in the requested resource-evidence scope.
Rust Foundation remains in progress on this exact head. This review does not replace the pending required checks. You are interacting with an AI system. |
|
@coderabbitai review exact current head Current exact-head authority is now terminal: Rust Foundation Please verify or refute this candidate source invariant before any mutation: in Treat termination, denial-of-service/resource bounds, subnormal transitions, overflow/underflow boundaries, midpoint/ties-to-even, or a possible direction reversal as actionable counterexamples. Do not recommend returning an unverified candidate, weakening coverage, adding private/artificial coverage fixtures, widening |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
src='crates/validation_core/src/bias_se.rs'
echo '== Exact identity =='
git rev-parse HEAD
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
echo '== Source map =='
ast-grep outline "$src" --items all --type function,struct,enum,impl
echo '== Exact rounder, helpers, and bounded route =='
git show "${head}:${src}" | nl -ba | sed -n '1,390p'
git show "${head}:${src}" | nl -ba | sed -n '390,680p'
echo '== All exact call sites and relevant tests =='
git grep -n -C 30 -E \
'\b(correctly_rounded_scaled_sqrt_ratio|compare_scaled_ratio_to_dyadic_square|adjacent_midpoint|neighbor_is_closer|positive_finite_dyadic|bias_standard_error|exact_pair_distance_standard_error)\s*\(' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Relevant full test files =='
git ls-tree -r --name-only "$head" -- crates/validation_core/tests |
grep -Ei '(bias.*(round|dyadic|midpoint|ratio|pair|standard)|standard.*error)' |
sort |
while IFS= read -r f; do
echo "----- $f -----"
git show "${head}:${f}" | nl -ba
done
echo '== History of correction loop and exact-comparison implementation =='
git log --reverse --format='%H %P %s' -- "$src"
git log -G 'for _ in 0\.\.4|neighbor_is_closer|correctly_rounded_scaled_sqrt_ratio' \
--format='%H %P %s' -- "$src"
git show --find-renames --unified=100 18822bc8744e09f2ccc4adaccd34ae40ba006e62 -- "$src"
echo '== Exact head check and artifact metadata =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}'
gh api "repos/${repo}/actions/artifacts/10164093741" \
--jq '{id,name,digest,expired,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha}}'
gh api "repos/${repo}/actions/artifacts/10164033659" \
--jq '{id,name,digest,expired,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha}}'Length of output: 2660 🏁 Script executed: #!/bin/bash
set -uo pipefail
repo='ContextualWisdomLab/TEPP'
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
src='crates/validation_core/src/bias_se.rs'
echo '== Source map =='
ast-grep outline "$src" --items all --type function,struct,enum || true
echo '== Exact rounder, helpers, and bounded route =='
git show "${head}:${src}" | nl -ba | sed -n '1,390p'
git show "${head}:${src}" | nl -ba | sed -n '390,680p'
echo '== All exact call sites and relevant tests =='
git grep -n -C 30 -E \
'\b(correctly_rounded_scaled_sqrt_ratio|compare_scaled_ratio_to_dyadic_square|adjacent_midpoint|neighbor_is_closer|positive_finite_dyadic|bias_standard_error|exact_pair_distance_standard_error)\s*\(' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Relevant full test files =='
git ls-tree -r --name-only "$head" -- crates/validation_core/tests |
grep -Ei '(bias.*(round|dyadic|midpoint|ratio|pair|standard)|standard.*error)' |
sort |
while IFS= read -r f; do
echo "----- $f -----"
git show "${head}:${f}" | nl -ba
done
echo '== History of correction loop and exact-comparison implementation =='
git log -G 'for _ in 0\.\.4|neighbor_is_closer|correctly_rounded_scaled_sqrt_ratio' \
--format='%H %P %s' -- "$src" || true
git show --find-renames --unified=100 18822bc8744e09f2ccc4adaccd34ae40ba006e62 -- "$src" || true
echo '== Exact head check and artifact metadata =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}' || true
gh api "repos/${repo}/actions/artifacts/10164093741" \
--jq '{id,name,digest,expired,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha}}' || true
gh api "repos/${repo}/actions/artifacts/10164033659" \
--jq '{id,name,digest,expired,workflow_run:{id:.workflow_run.id,head_sha:.workflow_run.head_sha}}' || trueLength of output: 119223
I found one actionable limitation in the proposed invariant. The invariant proves directionality after a valid correction step.
However, monotonicity over a finite binary64 set does not provide an acceptable resource bound. The set contains far too many ordered finite values to justify an unbounded correction loop. The proposed argument does not prove a maximum distance between the initial seed ((numerator as f64) / (denominator as f64)).sqrt() * unitand the correctly rounded target interval. It must account for numerator conversion, division, square root, final scaling, normal/subnormal transitions, and the zero or infinity seed exits. Without that bound, replacing Keep the four-pass I found no direction reversal, midpoint, or boundary-handling defect in the current capped loop. The exact-head checks confirm that quality, repository/Python, Live PostgreSQL, documentation, and resource-evidence jobs passed. The production line and branch coverage gate remains RED. You are interacting with an AI system. |
|
@coderabbitai review 현재 exact head 현재 admission은
Normal finite restoration에서 exact power-of-two multiplication은 exponent shift이므로 representable 범위 안에서는 추가 rounding이 없고, 이 bound는 한 binade에서 If this boundary analysis is correct, any finite correctly-roundable target needs at most two |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
src='crates/validation_core/src/bias_se.rs'
echo '== Exact revision identity =='
git show -s --format='head=%H%nparents=%P%nsubject=%s' "$head"
git rev-parse HEAD
echo '== bias_se.rs structure =='
ast-grep outline "$src" --items all --type function,struct,enum,impl
echo '== Exact rounder and helper implementation =='
git show "${head}:${src}" | nl -ba | sed -n '1,340p'
echo '== Exact bounded-route implementation and callers =='
git show "${head}:${src}" | nl -ba | sed -n '340,680p'
git grep -n -C 35 -E \
'\b(correctly_rounded_scaled_sqrt_ratio|exact_pair_distance_standard_error|exact_power_of_two|bias_standard_error)\s*\(' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Existing boundary and rounding contracts =='
git ls-tree -r --name-only "$head" -- crates/validation_core/tests |
grep -Ei '(bias.*(round|dyadic|midpoint|ratio|subnormal|overflow|boundary)|standard.*error)' |
sort |
while IFS= read -r f; do
echo "----- $f -----"
git show "${head}:${f}" | nl -ba
done
echo '== Exact loop introduction and changes =='
git log -G 'for _ in 0\.\.4|correctly_rounded_scaled_sqrt_ratio|neighbor_is_closer' \
--format='%H %P %s' -- "$src" || trueLength of output: 1906 🤖 web_search executed:
Length of output: 8359 🤖 web_search executed:
Length of output: 13484 🏁 Script executed: #!/bin/bash
set -euo pipefail
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
src='crates/validation_core/src/bias_se.rs'
echo '== Exact rounder and helpers: Lines 1-340 =='
git show "${head}:${src}" | nl -ba | sed -n '1,340p'
echo '== Exact public route and pair-distance construction: Lines 340-700 =='
git show "${head}:${src}" | nl -ba | sed -n '340,700p'
echo '== Definitions and all repository call sites =='
git grep -n -C 28 -E \
'\b(correctly_rounded_scaled_sqrt_ratio|exact_pair_distance_standard_error|exact_power_of_two|neighbor_is_closer|adjacent_midpoint|compare_scaled_ratio_to_dyadic_square)\s*\(' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Exact tests that exercise rounding or binary64 boundaries =='
git ls-tree -r --name-only "$head" -- crates/validation_core/tests |
grep -Ei '(bias.*(round|dyadic|midpoint|ratio|subnormal|overflow|boundary)|standard.*error)' |
sort |
while IFS= read -r f; do
echo "----- $f -----"
git show "${head}:${f}" | nl -ba
done
echo '== Arithmetic operations and finite/zero/infinite exits in exact source =='
git show "${head}:${src}" | nl -ba | \
grep -E -C 8 'as f64|\.sqrt\(\)|/| \* |is_finite|== 0\.0|MIN_POSITIVE|MAX|to_bits|next_after|neighbor' || true
echo '== Current exact head status =='
git show -s --format='head=%H%nsubject=%s' "$head"
git rev-parse HEADLength of output: 118296 🏁 Script executed: #!/bin/bash
set -euo pipefail
head='a68055f108fc30b9b56348e8a5381ab436c813b3'
src='crates/validation_core/src/bias_se.rs'
echo '== exact_power_of_two and seed-entry region =='
git show "${head}:${src}" | nl -ba | sed -n '200,310p'
echo '== All direct internal calls to the private rounder at exact head =='
git grep -n -C 12 'correctly_rounded_scaled_sqrt_ratio(' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs'
echo '== Production-derived bounds before the rounder call =='
git show "${head}:${src}" | nl -ba | sed -n '391,455p'
echo '== Rust toolchain version selected by this revision =='
git ls-tree -r --name-only "$head" | grep -E '(^|/)(rust-toolchain(\.toml)?|Cargo\.toml)$' | while IFS= read -r f; do
echo "----- $f -----"
git show "${head}:${f}"
doneLength of output: 65030 🤖 web_search executed:
Length of output: 9676
정상 finite restoration에서는 입니다. 한 binade의 최소 local spacing으로 환산한 최악 경계는 normal-to-subnormal restoration도 반증하지 못했습니다. subnormal 구간에서는 restoration 전의 상대 오차가 최대값 근처에서도 한 minimum-subnormal quantum보다 작습니다. 최종 multiplication의 nearest-even 반올림과 target의 correctly rounded 반올림은 각각 최대 반 quantum을 더합니다. 따라서 seed와 올바른 nonzero subnormal 결과의 거리는 두 adjacent 값 이하입니다. 다음 경계도 확인했습니다.
따라서 다음과 같이 범위를 정확히 구분해야 합니다.
이 제한을 명시하면, 현재 Rust 1.98.0의 toolchain 지정과 You are interacting with an AI system. |
|
@coderabbitai review Please review exact head |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='0bd9d9eeb9f3d6be27548f46b93e597142f67668'
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
echo '== Exact commit identity =='
git rev-parse HEAD
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
echo '== Changed paths and focused diff =='
git diff --name-status "$parent" "$head"
git diff --find-renames --unified=140 "$parent" "$head" -- \
'crates/validation_core/tests/*.rs' \
crates/validation_core/src/bias.rs \
crates/validation_core/src/bias_se.rs \
Cargo.toml
echo '== Added or modified large-sample characterization contracts =='
git diff --name-only "$parent" "$head" -- 'crates/validation_core/tests/*.rs' |
while IFS= read -r file; do
echo "----- $file -----"
ast-grep outline "$file" --items all --type function,struct,enum,impl 2>/dev/null || true
cat -n "$file"
done
echo '== Public route and n=3..=16 exact-admission boundary =='
git show "${head}:crates/validation_core/src/bias_se.rs" | nl -ba | sed -n '330,680p'
git show "${head}:crates/validation_core/src/bias.rs" | nl -ba | sed -n '1,220p'
git grep -n -C 22 -E \
'\b(pub )?fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error|exact_pair_distance_standard_error' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Related public two-level, permutation, subnormal, and false-zero contracts =='
git ls-tree -r --name-only "$head" -- crates/validation_core/tests |
grep -Ei '(bias.*(large|reference|two.level|permutation|subnormal|false.zero)|standard.*error)' |
sort |
while IFS= read -r file; do
echo "----- $file -----"
git show "${head}:${file}" | nl -ba | grep -E -C 10 \
'(2047|2_047|two.level|anchor|permut|MIN|subnormal|false.zero|bias_standard_error|to_bits|expect)' || true
done
echo '== Exact-head GitHub state =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}' || true
gh pr view 488 --repo "$repo" --json headRefOid,isDraft,state,mergeStateStatus,statusCheckRollup \
--jq '{headRefOid,isDraft,state,mergeStateStatus,checks:[.statusCheckRollup[]|{name,status,conclusion,detailsUrl}]}' || trueLength of output: 125950 You are interacting with an AI system. |
|
@coderabbitai review Exact head |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='aaf7921a657bc55d0b346943599439839fa4e4e3'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
echo '== Exact head =='
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
git rev-parse HEAD
echo '== Requested test file =='
git show "${head}:${file}" | nl -ba
echo '== Exact-head diff for requested test file =='
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
git diff --find-renames --unified=120 "$parent" "$head" -- "$file"
echo '== Public route and exact-admission boundary =='
git show "${head}:crates/validation_core/src/bias_se.rs" | nl -ba | sed -n '360,610p'
git grep -n -C 20 -E \
'\b(pub )?fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error|exact_pair_distance_standard_error' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Exact-head checks and Rust Foundation jobs =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {id,name,status,conclusion,details_url,completed_at}'
gh api "repos/${repo}/actions/runs?head_sha=${head}&per_page=100" \
--jq '.workflow_runs[] | {id,name,path,status,conclusion,head_sha,html_url}' \
> /tmp/exact-head-runs.json
cat /tmp/exact-head-runs.json
echo '== Rust Foundation formatter logs =='
jq -r '. | select(.name == "Rust Foundation") | .id' /tmp/exact-head-runs.json | while IFS= read -r run_id; do
[ -n "$run_id" ] || continue
gh api "repos/${repo}/actions/runs/${run_id}/jobs?per_page=100" \
--jq '.jobs[] | {id,name,status,conclusion,html_url,steps:[.steps[]|{number,name,conclusion}]}' \
> "/tmp/rust-jobs-${run_id}.json"
cat "/tmp/rust-jobs-${run_id}.json"
jq -r '.id' "/tmp/rust-jobs-${run_id}.json" | while IFS= read -r job_id; do
[ -n "$job_id" ] || continue
echo "----- job ${job_id} formatter excerpts -----"
gh api "repos/${repo}/actions/jobs/${job_id}/logs" 2>/dev/null | \
grep -E -C 20 'cargo fmt|rustfmt|bias_large_sample_reference_characterization_contract\.rs|Diff in ' || true
done
doneLength of output: 72861 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
run='34514509582'
job='102996525486'
head='aaf7921a657bc55d0b346943599439839fa4e4e3'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
echo '== Failed job metadata and steps =='
gh api "repos/${repo}/actions/jobs/${job}" \
--jq '{id,name,status,conclusion,started_at,completed_at,html_url,steps:[.steps[]|{number,name,status,conclusion,started_at,completed_at}]}'
echo '== Failed-job logs: gh run view =='
gh run view "$run" --repo "$repo" --log-failed 2>&1 || true
echo '== Failed-job logs: REST redirect endpoint =='
curl -fsSL \
-H "Authorization: Bearer $(gh auth token)" \
-H "Accept: application/vnd.github+json" \
"https://api.github.com/repos/${repo}/actions/jobs/${job}/logs" \
2>&1 | grep -E -C 25 \
'cargo fmt|rustfmt|Diff in |bias_large_sample_reference_characterization_contract\.rs|error:' || true
echo '== Formatter configuration at exact head =='
git ls-tree -r --name-only "$head" | grep -E '(^|/)(rustfmt\.toml|\.rustfmt\.toml|rust-toolchain(\.toml)?)$' | while IFS= read -r config; do
echo "----- $config -----"
git show "${head}:${config}"
done
echo '== Candidate long lines in requested test =='
git show "${head}:${file}" | awk 'length($0) > 100 { printf "%d:%d:%s\n", NR, length($0), $0 }'
echo '== Current Rust Foundation run state =='
gh api "repos/${repo}/actions/runs/${run}" \
--jq '{id,status,conclusion,head_sha,updated_at,html_url}'Length of output: 16405
The pinned Rust 1.98.0 -use validation_core::{bias_standard_error, ValidationError};
+use validation_core::{ValidationError, bias_standard_error};- let second = bias_standard_error(&truth, &recovered)
- .expect("deterministic repeated result");
+ let second =
+ bias_standard_error(&truth, &recovered).expect("deterministic repeated result");The job failed at I found no scientific-scope issue in the test source.
The semantics are correct by source review. CI has not executed this test on the exact head because formatting stopped the quality job. You are interacting with an AI system. |
|
@coderabbitai review Please review exact current head |
|
@coderabbitai review Exact head |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='40582fb4a01c8f436ee7383a7b909f50f03ea6ee'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
echo '== exact identity =='
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
git rev-parse HEAD
echo '== changed paths and requested diff =='
git diff --name-status "$parent" "$head"
git diff --find-renames --unified=120 "$parent" "$head" -- "$file"
echo '== requested test source =='
git show "${head}:${file}" | nl -ba
echo '== public route and fallback implementation =='
git show "${head}:crates/validation_core/src/bias_se.rs" | nl -ba | sed -n '360,650p'
git show "${head}:crates/validation_core/src/bias.rs" | nl -ba | sed -n '1,190p'
git grep -n -C 22 -E \
'\b(pub )?fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error|exact_pair_distance_standard_error|standard_error_from_deviations' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== relevant error definitions =='
git grep -n -C 10 -E 'enum ValidationError|InvalidInput' \
"$head" -- 'crates/validation_core/src/*.rs' || true
echo '== independent exact rounding check =='
python3 - <<'PY'
from fractions import Fraction
n = 2047
p = (1 << 52) + 1
expected = 0x4280_0200_4008_0101
# Decode a positive finite binary64 bit pattern to an exact rational.
def decode(bits):
exponent = (bits >> 52) & 0x7ff
fraction = bits & ((1 << 52) - 1)
if exponent == 0:
return Fraction(fraction, 1) * Fraction(2) ** -1074
return Fraction((1 << 52) | fraction, 1) * Fraction(2) ** (exponent - 1023 - 52)
target = Fraction(p, n)
candidate = decode(expected)
below = decode(expected - 1)
above = decode(expected + 1)
mid_lower = (below + candidate) / 2
mid_upper = (candidate + above) / 2
print(f'target={p}/{n}')
print(f'candidate_bits=0x{expected:016x}')
print(f'candidate={candidate.numerator}/{candidate.denominator}')
print(f'target_in_candidate_interval={mid_lower < target < mid_upper}')
print(f'distance_to_candidate={abs(target-candidate)}')
print(f'distance_to_below={abs(target-below)}')
print(f'distance_to_above={abs(target-above)}')
print(f'nearest_unique={abs(target-candidate) < abs(target-below) and abs(target-candidate) < abs(target-above)}')
PY
echo '== exact-head check state =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}' || true
gh pr view 488 --repo "$repo" --json headRefOid,isDraft,state,mergeStateStatus,statusCheckRollup \
--jq '{headRefOid,isDraft,state,mergeStateStatus,checks:[.statusCheckRollup[]|{name,status,conclusion,detailsUrl}]}' || trueLength of output: 203354 🏁 Script executed: #!/bin/bash
set -euo pipefail
log='/tmp/coderabbit-shell-logs/shell-output-CCnOCC'
echo '== log section markers =='
rg -n '^== ' "$log"
echo '== exact identity, diff, and requested test =='
sed -n '1,260p' "$log"
echo '== public route section markers and bounded excerpts =='
rg -n -C 10 \
'pub fn bias_standard_error|exact_pair_distance_standard_error|3\.\.=16|crate::bias::bias_standard_error|SAMPLE_COUNT|anchor|minimum_subnormal' \
"$log" | head -n 700
echo '== failed quality job diagnostic =='
rg -n -C 30 \
'Format, lint, test, rustdoc, and dependency policy|cargo fmt|rustfmt|Diff in |error:' \
"$log" | tail -n 500Length of output: 119879 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
run='34514810357'
job='102997534915'
echo '== job metadata =='
gh api "repos/${repo}/actions/jobs/${job}" \
--jq '{id,name,status,conclusion,html_url,steps:[.steps[]|{number,name,status,conclusion}]}'
echo '== failed-job log excerpts =='
gh run view "$run" --repo "$repo" --log-failed 2>&1 | \
rg -n -C 25 \
'Check formatting|cargo fmt|rustfmt|Diff in |error(\[|:)|failed|bias_large_sample_reference_characterization_contract' || trueLength of output: 2342 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
job='102997534915'
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
gh api "repos/${repo}/actions/jobs/${job}/logs" > "${tmpdir}/job-log.zip"
unzip -Z1 "${tmpdir}/job-log.zip" | head -n 30
unzip -p "${tmpdir}/job-log.zip" | rg -n -C 30 \
'Run Clippy|error(\[|:)|warning:|clippy::|bias_large_sample_reference_characterization_contract|could not compile' || trueLength of output: 835 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
job='102997534915'
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
curl -fsSL \
-H "Authorization: Bearer $(gh auth token)" \
-H 'Accept: application/vnd.github+json' \
"https://api.github.com/repos/${repo}/actions/jobs/${job}/logs" \
-o "${tmpdir}/job-log.zip"
file "${tmpdir}/job-log.zip"
unzip -Z1 "${tmpdir}/job-log.zip" | head -n 20
unzip -p "${tmpdir}/job-log.zip" |
sed $'s/\x1b\\[[0-9;]*[[:alpha:]]//g' |
rg -n -C 25 \
'Run Clippy|error(\[|:)|warning:|clippy::|bias_large_sample_reference_characterization_contract|could not compile' || trueLength of output: 1028 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
job='102997534915'
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
curl -fsSL \
-H "Authorization: Bearer $(gh auth token)" \
-H 'Accept: application/vnd.github+json' \
"https://api.github.com/repos/${repo}/actions/jobs/${job}/logs" \
-o "${tmpdir}/job-log.zip"
unzip -Z1 "${tmpdir}/job-log.zip" | head -n 20
unzip -p "${tmpdir}/job-log.zip" |
sed $'s/\x1b\\[[0-9;]*[[:alpha:]]//g' |
rg -n -C 25 \
'Run Clippy|error(\[|:)|warning:|clippy::|bias_large_sample_reference_characterization_contract|could not compile' || trueLength of output: 1394
I found no scientific, determinism, or test-oracle defect in the requested scope.
Exact-head verification remains incomplete. Formatting passed, but Clippy failed, so the test suite did not run. The production coverage job was also still in progress. I could not retrieve the Clippy diagnostic from the Actions log archive in this environment. You are interacting with an AI system. |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='40582fb4a01c8f436ee7383a7b909f50f03ea6ee'
job='102997534915'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
echo '== Exact revision =='
git show -s --format='head=%H%nparents=%P%nsubject=%s' "$head"
git rev-parse HEAD
echo '== Failed job metadata =='
gh api "repos/${repo}/actions/jobs/${job}" \
--jq '{id,name,status,conclusion,head_sha,html_url,steps:[.steps[]|{number,name,status,conclusion}]}'
echo '== Failed Clippy diagnostic =='
gh api "repos/${repo}/actions/jobs/${job}/logs" |
grep -E -C 20 \
'cargo clippy|bias_large_sample_reference_characterization_contract\.rs|warning:|error:|help:|note:' \
|| true
echo '== Exact affected test source =='
git show "${head}:${file}" | nl -ba
echo '== Exact-head change to affected test =='
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
git diff --find-renames --unified=100 "$parent" "$head" -- "$file"
echo '== Workspace lint configuration =='
for f in Cargo.toml crates/validation_core/Cargo.toml; do
if git cat-file -e "${head}:${f}" 2>/dev/null; then
echo "----- $f -----"
git show "${head}:${f}"
fi
doneLength of output: 15313 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
run='34514810357'
job='102997534915'
file='bias_large_sample_reference_characterization_contract.rs'
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
echo '== Failed-step log through gh run view =='
gh run view "$run" --repo "$repo" --log-failed 2>&1 |
sed -E $'s/\x1B\\[[0-?]*[ -\\/]*[`@-`~]//g' |
grep -E -C 25 \
'Run Clippy|cargo clippy|'"$file"'|warning:|error:|help:|note:|clippy::' \
|| true
echo '== Raw job log through REST redirect =='
curl -fsSL \
-H "Authorization: Bearer $(gh auth token)" \
-H 'Accept: application/vnd.github+json' \
"https://api.github.com/repos/${repo}/actions/jobs/${job}/logs" \
-o "${tmpdir}/job.log"
sed -E $'s/\x1B\\[[0-?]*[ -\\/]*[`@-`~]//g' "${tmpdir}/job.log" |
grep -E -C 25 \
'Run Clippy|cargo clippy|'"$file"'|warning:|error:|help:|note:|clippy::' \
|| true
echo '== Unstripped matching raw lines for source coordinates =='
grep -a -n -E \
"$file|clippy::|error:|warning:" \
"${tmpdir}/job.log" \
|| trueLength of output: 10627
Rust Foundation job The exact diagnostic is: Use an exact binary64 construction. This is the minimal semantics-preserving fix: let gap = f64::from_bits(0x4330_0000_0000_0001);This bit pattern is exactly
Failed to handle agent chat message. Please try again. You are interacting with an AI system. |
|
@coderabbitai review Please review exact current head |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='01ca3088b4a4bc514a441c4ab8ca4430219ab9e9'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
echo '== Exact revision =='
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
git rev-parse HEAD
echo '== Requested test source =='
git show "${head}:${file}" | nl -ba
echo '== Requested test diff =='
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
git diff --find-renames --unified=120 "$parent" "$head" -- "$file"
echo '== Public admission and fallback routing =='
git show "${head}:crates/validation_core/src/bias_se.rs" | nl -ba | sed -n '380,500p'
git grep -n -C 18 -E \
'\bpub fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error|exact_pair_distance_standard_error' \
"$head" -- 'crates/validation_core/src/*.rs' 'crates/validation_core/tests/*.rs' || true
echo '== Fallback false-zero behavior and error definition =='
git grep -n -C 16 -E \
'\bfn bias_standard_error\b|standard_error == 0\.0|InvalidInput|enum ValidationError' \
"$head" -- crates/validation_core/src/bias.rs crates/validation_core/src/lib.rs crates/validation_core/src/error.rs || true
echo '== Exact rational checks =='
python3 - <<'PY'
from fractions import Fraction
N = 2047
gap = (1 << 52) + 1
expected_bits = 0x4280_0200_4008_0101
gap_bits = 0x4330_0000_0000_0001
def decode(bits):
sign = -1 if bits >> 63 else 1
exponent = (bits >> 52) & 0x7ff
fraction = bits & ((1 << 52) - 1)
if exponent == 0:
return sign * Fraction(fraction, 1) * Fraction(2) ** -1074
assert exponent != 0x7ff
return sign * Fraction((1 << 52) | fraction, 1) * Fraction(2) ** (exponent - 1023 - 52)
target = Fraction(gap, N)
candidate = decode(expected_bits)
below = decode(expected_bits - 1)
above = decode(expected_bits + 1)
print(f'gap_literal_equals_2^52_plus_1={decode(gap_bits) == gap}')
print(f'two_level_se_identity_target={gap}/{N}')
print(f'expected_bits=0x{expected_bits:016x}')
print(f'unique_nearest={abs(target-candidate) < abs(target-below) and abs(target-candidate) < abs(target-above)}')
print(f'not_a_tie={2*abs(target-candidate) != abs(above-candidate) and 2*abs(target-candidate) != abs(candidate-below)}')
print(f'min_subnormal_target_below_half_quantum={Fraction(1, N) < Fraction(1, 2)}')
PY
echo '== Exact-head checks =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}' || trueLength of output: 123868
I found no scientific or determinism defect in the requested scope.
The exact-head format, lint, test, rustdoc, dependency policy, repository/Python, PostgreSQL, resource-evidence, and governance checks are GREEN. The Production line and branch coverage check is RED. You are interacting with an AI system. |
|
@coderabbitai review Exact head |
|
Tip For best results, initiate chat on the files or code changes. 🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='1b50af75a4e2449922252f805552860fbb0494dc'
file='crates/validation_core/tests/bias_large_sample_reference_characterization_contract.rs'
parent="$(git show -s --format=%P "$head" | awk '{print $1}')"
echo '== exact revision =='
git show -s --format='head=%H%nparents=%P%nsubject=%s%ncommit_date=%cI' "$head"
git rev-parse HEAD
echo '== focused delta =='
git diff --find-renames --unified=160 "$parent" "$head" -- "$file"
echo '== full requested test =='
git show "${head}:${file}" | nl -ba
echo '== public API routing and admission boundary =='
git show "${head}:crates/validation_core/src/bias_se.rs" | nl -ba | sed -n '370,500p'
git grep -n -C 18 -E \
'\bpub fn bias_standard_error\b|3\.\.=16|n\s*<\s*3|n\s*>\s*16|crate::bias::bias_standard_error|exact_pair_distance_standard_error' \
"$head" -- 'crates/validation_core/src/*.rs'
echo '== exact binary64 validation for stated specimens =='
python3 - <<'PY'
from fractions import Fraction
import struct
gap = (1 << 52) + 1
cases = [(17, 5), (63, 21), (255, 85), (1023, 341)]
def decode(bits):
sign = -1 if bits >> 63 else 1
exp = (bits >> 52) & 0x7ff
frac = bits & ((1 << 52) - 1)
if exp == 0:
return sign * Fraction(frac, 1) * Fraction(2) ** -1074
assert exp != 0x7ff
return sign * Fraction((1 << 52) | frac, 1) * Fraction(2) ** (exp - 1023 - 52)
def float_bits(value):
return struct.unpack('>Q', struct.pack('>d', value))[0]
for n, m in cases:
squared = Fraction(gap * gap * m * (n - m), n * n * (n - 1))
# Python float sqrt is only used to identify candidate; exact neighbor proof follows.
candidate_bits = float_bits(float(squared) ** 0.5)
candidate = decode(candidate_bits)
below = decode(candidate_bits - 1)
above = decode(candidate_bits + 1)
# Compare squared distances to avoid irrational arithmetic:
# target root nearest interval is bounded by midpoints; test candidate^2 relative to target.
lower_mid_sq = ((below + candidate) / 2) ** 2
upper_mid_sq = ((candidate + above) / 2) ** 2
correct = lower_mid_sq < squared < upper_mid_sq
print(
f'n={n} m={m} expected_bits=0x{candidate_bits:016x} '
f'unique_nearest={correct} '
f'formula_numerator={gap * gap * m * (n-m)} '
f'formula_denominator={n*n*(n-1)}'
)
PY
echo '== exact-head checks =='
gh api "repos/${repo}/commits/${head}/check-runs?per_page=100" \
--jq '.check_runs[] | {name,status,conclusion,details_url,completed_at}' || trueLength of output: 59038 🏁 Script executed: #!/bin/bash
set -euo pipefail
repo='ContextualWisdomLab/TEPP'
head='1b50af75a4e2449922252f805552860fbb0494dc'
job='103004914630'
echo '== exact asserted-literal checks =='
python3 - <<'PY'
from fractions import Fraction
gap = (1 << 52) + 1
cases = [
(17, 5, 0x42fd_294a_104a_a492),
(63, 21, 0x42ee_a713_4c43_1b35),
(255, 85, 0x42de_49d6_e90b_6ea0),
(1023, 341, 0x42ce_330b_92b8_7e1b),
]
def decode(bits):
exponent = (bits >> 52) & 0x7ff
fraction = bits & ((1 << 52) - 1)
assert 0 < exponent < 0x7ff
return Fraction((1 << 52) | fraction, 1) * Fraction(2) ** (exponent - 1023 - 52)
for n, m, bits in cases:
target_sq = Fraction(gap * gap * m * (n - m), n * n * (n - 1))
lower, value, upper = decode(bits - 1), decode(bits), decode(bits + 1)
lower_boundary_sq = ((lower + value) / 2) ** 2
upper_boundary_sq = ((value + upper) / 2) ** 2
print(f'n={n} m={m} bits=0x{bits:016x}')
print(f' unique_nearest={lower_boundary_sq < target_sq < upper_boundary_sq}')
print(f' lower_margin_sign={(target_sq - lower_boundary_sq) > 0}')
print(f' upper_margin_sign={(upper_boundary_sq - target_sq) > 0}')
print(f' lower_neighbor_is_not_nearer={target_sq > lower_boundary_sq}')
print(f' upper_neighbor_is_not_nearer={target_sq < upper_boundary_sq}')
PY
echo '== quality job failure diagnostic =='
tmpdir="$(mktemp -d)"
trap 'rm -rf "$tmpdir"' EXIT
curl -fsSL \
-H "Authorization: Bearer $(gh auth token)" \
-H 'Accept: application/vnd.github+json' \
"https://api.github.com/repos/${repo}/actions/jobs/${job}/logs" \
-o "${tmpdir}/job.log"
sed -E $'s/\x1B\\[[0-?]*[ -\\/]*[`@-`~]//g' "${tmpdir}/job.log" | \
rg -n -C 20 \
'cargo fmt|cargo clippy|error(\[|:)|warning:|clippy::|bias_large_sample_reference_characterization_contract|test result' \
|| true
echo '== exact PR state =='
gh pr view 488 --repo "$repo" --json headRefOid,isDraft,mergeStateStatus,statusCheckRollup \
--jq '{headRefOid,isDraft,mergeStateStatus,checks:[.statusCheckRollup[]|{name,status,conclusion,detailsUrl}]}' || trueLength of output: 14931
The analytic identity is correct: [ Exact rational midpoint checks confirm that each hard-coded value is the unique nearest binary64 result for this identity. This includes However, the exact-head Rust Foundation job ran the test and reported: The failure occurs on the first new Therefore, the new test does not currently characterize the established fallback correctly. It instead asserts a mathematical nearest-result oracle that the fallback does not produce for this specimen. The test also stops before it can establish prefix/suffix/rotation equivalence or repeat determinism for the remaining new specimens. The delta does not widen the You are interacting with an AI system. |
1 participant
Validation Evidence authority
#488 remains the owner-correct Validation Evidence landing vehicle, stacked on prerequisite
#492@794ba9e6dda9f043aa499920fdf609b81b075d7e; protected release authority remainsmain@a243f18da4a4ca8a8d068c39922537f1f8ed6ad0. Current exact head is01ca3088b4a4bc514a441c4ab8ca4430219ab9e9. The PR remains Draft/open/mergeable. No force update, destructive rebase, self-approval, gate weakening, coverage exclusion, mutable sibling dependency, copied provider routing, or copied generalized psychometric owner logic is used.Production
validation_core::bias_standard_errorstill keeps bounded represented-input exact admission forn=3..=16;n=2retains its direct identity andn>16remains characterization-only under #491. Reusable generalized/static psychometric arithmetic remains fast-mlsirm-owned. Semantic LLM routing remains contextual-orchestrator-owned through an immutable released contract only.New large-sample reference characterization
Exact head
01ca3088...addscrates/validation_core/tests/bias_large_sample_reference_characterization_contract.rsas a public fallback reference baseline for future #491 admission work. It does not widen production admission and is not an independent scientific oracle.The fixture uses
n=2,047with one zero residual and 2,046 identical nonzero residuals. Forgap = f64::from_bits(0x4330_0000_0000_0001) = 2^52+1, the exact two-level identity isSE(mean)=|gap|/2047; the nearest binary64 is asserted bitwise as0x4280_0200_4008_0101. The zero anchor is placed at indices0,1_023, and2_046, and each case is evaluated twice to preserve permutation and repeat-determinism evidence. A second specimen usesgap=2^-1074and requires publicInvalidInput, preserving false-zero refusal because the exact positive SE is below the half-minimum-subnormal boundary.A focused CodeRabbit review on the exact current head found no scientific or determinism defect in this scope. It independently confirmed the binary64 literal, exact two-level identity, unique nearest expected result (not a midpoint tie), anchor-position permutation specimens, minimum-subnormal refusal, and that
n=2,047continues to delegate to the established fallback rather than widening then=3..=16exact-proof route. Technical bot review does not substitute for the organization-required approving GitHub review.RED -> repair -> exact-head GREEN evidence
The new test initially exposed quality-gate REDs rather than being allowed to bypass them. The formatting mismatch was repaired until
cargo fmt --all -- --checkpassed. The next exact-head run then failed Clippy; the only scientific expression changed afterward was the generic integer-to-f64construction of2^52+1, replaced by its exact binary64 bit patternf64::from_bits(0x4330_0000_0000_0001). On01ca3088..., formatting and Clippy are both GREEN, establishing the causal repair without suppressing lints.Exact-head run authority on
01ca3088...:34515053132: GREEN.34515053109: GREEN. Immutable artifact10167332804,bias-se-proof-budget-01ca3088b4a4bc514a441c4ab8ca4430219ab9e9, digestsha256:eb274cedd6b3b40d4a558755af900e42a0f4869b754252faf0d37be645f25acc.34515053075: RED only in the owned-production 100% line/branch coverage job. Live PostgreSQL, repository/Python contracts plus SBOM/provenance, rustfmt, Clippy, no-retry test suite, doctests, warning-free rustdoc, and dependency/license/advisory/source policy are GREEN on the same exact head.10167422485(sha256:b9043630f0dd6e87b35bb2d8fd296e20593b551b72d9ac767b2146a4b70fc107) and branch10167481107(sha256:10b3ef144a15c6e2e998a6cdb21dc3f0709fe4d8760f30ee622a4e7da298e2b8). Both 100% enforcement steps remain RED. Do not claim a coverage reduction from this test until those exact artifact payloads are folded underscripts/check_coverage.pyauthority.The last fully parsed production coverage frontier remains the predecessor source authority: 4,449/4,452 branch arms with 3 missing outcomes, all in
bias.rs(68-69positive-SE false-zero refusal,148target-usizerational-denominator guard,589-590translated nonpositive-dispersion refusal), plusbias_se.rs:307as a separate unexecuted terminal four-step correction exhaustion line rather than a branch miss. This head changes tests only, not production arithmetic or the coverage denominator; nevertheless exact-current artifact counts are not promoted without parsing the current artifacts.Resource evidence and #491 boundary
The isolated
n=2,047,diameter=2^58+1resource harness remains characterization evidence. Candidate and fallback run in separate/usr/bin/time -vprocesses on the identical deterministic geometry, preventing process-RSS attribution leakage. The previously verified specimen records candidate p9523,306 ns, max RSS2,176 kB, zero pair-record scratch versus fallback p9515,385,815 ns, max RSS100,152 kB,2,094,081pair records /100,515,888 bytesexplicit scratch. These are specimen-specific kernel/process measurements, not a portable performance guarantee and not buyer-facing HTTP p95.The new reference contract closes part of #491's evidence gap for one
n=2,047two-level geometry: bitwise represented-result identity, three anchor-position permutation specimens, repeated determinism, and minimum-subnormal refusal. It does not justify generaln>16admission. Wider production admission still requires a defined admitted domain, deterministic/reference/refusal/permutation evidence over that domain, applicable release CPU/allocator/RSS and buyer-path latency evidence, exact-current-head 100% authored production line+branch coverage, qualifying independent current-head review, code-current TRACEABILITY/product-technical-gap authority, normal #492 prerequisite landing, and immutable released owner contracts.bias_se.rs:307also remains a bounded-resource refusal. Candidate/midpoint comparisons are total, but no complete proof yet bounds the actual rounded seed to the correct binary64 interval within a small number of adjacent corrections across integer conversion, division, sqrt, power-of-two restoration, subnormal transition, and maximum-finite boundary. Do not replace the fixed cap with an unbounded loop or return an unverified candidate.Review / foundation boundary
Organization ruleset
18156473requires qualifying approval, stale-review dismissal after pushes, review-thread resolution, central required workflows, and non-fast-forward protection. Administrative bypass is not delivery evidence and is not used. #492 remains intentionally Draft at exact794ba9e6dda9f043aa499920fdf609b81b075d7e;.github#2038and contextual-orchestrator #1023/#1083 remain the canonical owner path for a releasedorchestrator/freeActions contract. TEPP must not copy unpublished provider routing/credentials/model selection or consume mutable contextual-orchestrator source as production authority.Before protected merge, require one exact current head with owned-production 100% authored line+branch coverage, current Rust/security/documentation/review gates, truthful route telemetry, code-current TRACEABILITY/product-technical-gap authority, normal foundation landing, immutable released dependency contracts, rollback/recovery/release evidence where applicable, and qualifying independent review.