Repository navigation
feat(audit): rebuild bounded snapshot manifests on current pagination - #97
seonghobae wants to merge 5 commits into
Conversation
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks 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 |
|
@coderabbitai review |
|
|
@coderabbitai review |
|
|
@coderabbitai review Please review the exact current stacked source head |
🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
base='f9062821556715537a028095326544bf234e45ae'
head='deed55787f4b7eb1ef57147071957f03631a3b1c'
git cat-file -e "${base}^{commit}"
git cat-file -e "${head}^{commit}"
printf '%s\n' '=== exact range ==='
git diff --stat "${base}" "${head}"
git diff --name-status "${base}" "${head}"
printf '%s\n' '=== changed snapshot-related files ==='
git diff --name-only "${base}" "${head}" | rg '(^|/)(checkpoint_audit|__init__|CHANGELOG|README|ADR|doctor|audit|snapshot|test).*' || true
printf '%s\n' '=== implementation outlines ==='
fd -a '^checkpoint_audit\.py$' . -E .git -x ast-grep outline '{}' --items all --view expanded
printf '%s\n' '=== focused range diff ==='
git diff --find-renames --find-copies --unified=80 "${base}" "${head}" -- \
'pg_llm_batch/checkpoint_audit.py' \
'pg_llm_batch/__init__.py' \
'CHANGELOG.md' \
':!*.lock' || trueLength of output: 44363 🏁 Script executed: #!/bin/bash
set -euo pipefail
printf '%s\n' '=== implementation: page and manifest methods ==='
sed -n '430,760p' pg_llm_batch/checkpoint_audit.py
printf '%s\n' '=== test structure ==='
for file in \
tests/test_checkpoint_audit_snapshot_manifest.py \
tests/test_checkpoint_audit_snapshot_transaction_boundary.py \
tests/test_checkpoint_audit_snapshot_digest_vector.py \
tests/test_checkpoint_audit_snapshot_integration.py \
tests/test_checkpoint_audit_snapshot_public_api.py \
tests/test_checkpoint_audit_ci_contract.py
do
printf '\n=== %s ===\n' "$file"
ast-grep outline "$file" --items all --view expanded
done
printf '%s\n' '=== assertions relevant to requested scope ==='
rg -n -C 4 -i \
'repeatable|serializable|read.?only|autocommit|isolation|transaction|snapshot|concurr|overflow|max_events|page_size|timestamp|recorded_at|digest|sha256|revalid|outside the requested key|cursor' \
tests/test_checkpoint_audit_snapshot_manifest.py \
tests/test_checkpoint_audit_snapshot_transaction_boundary.py \
tests/test_checkpoint_audit_snapshot_digest_vector.py \
tests/test_checkpoint_audit_snapshot_integration.py \
tests/test_checkpoint_audit_snapshot_public_api.py \
tests/test_checkpoint_audit_ci_contract.py
printf '%s\n' '=== documentation contract ==='
rg -n -C 5 -i \
'build_audit_snapshot_manifest|snapshot manifest|repeatable|serializable|read.?only|autocommit|isolation|100,000|digest|timestamp|sha-256' \
CHANGELOG.md docs/checkpoint-audit.md \
docs/adr/0012-checkpoint-audit-snapshot-manifests.md \
docs/doctoring/checkpoint-audit-snapshot-manifests.md \
ARCHITECTURE.md AGENTS.md CLAUDE.md
printf '%s\n' '=== CI additions ==='
git diff --unified=50 \
f9062821556715537a028095326544bf234e45ae \
deed55787f4b7eb1ef57147071957f03631a3b1c \
-- .github/workflows/ci.yml tests/test_checkpoint_audit_ci_contract.pyLength of output: 209297
The implementation enforces an active, read-only The manifest reuses the predecessor keyset page method. Exact audit-key revalidation, strict descending identities, cursor checks, and bounded lookahead remain in effect. The traversal remains bounded. It uses at most one page plus fixed digest state. It fails closed when the event budget is exceeded. The digest framing includes the domain, trusted key, every retained event field, UTC microsecond timestamps, and final count and identity bounds. It excludes page partitioning. The fixed digest vector and timestamp-offset tests support this contract. The range includes deterministic and live concurrent-insert coverage. CI invokes the live snapshot integration test. The package exports and public documentation match the implemented transaction, isolation, overflow, and SHA-256 boundaries. This result is limited to staged Draft source review of the specified range. It does not treat predecessor, default-branch, or dependency-order evidence as acceptance evidence.
|
|
@coderabbitai review Please review exact current replacement head |
🧩 Analysis chain🏁 Script executed: #!/bin/bash
set -euo pipefail
base='f9062821556715537a028095326544bf234e45ae'
head='deed55787f4b7eb1ef57147071957f03631a3b1c'
git cat-file -e "${base}^{commit}"
git cat-file -e "${head}^{commit}"
printf '%s\n' '=== exact range ==='
git diff --stat "$base" "$head"
git diff --name-status "$base" "$head"
printf '%s\n' '=== changed Python module structure ==='
ast-grep outline pg_llm_batch/checkpoint_audit.py --items all --view expanded
printf '%s\n' '=== changed public exports structure ==='
ast-grep outline pg_llm_batch/__init__.py --items all --view expanded
printf '%s\n' '=== focused implementation diff ==='
git diff --find-renames --unified=100 "$base" "$head" -- \
pg_llm_batch/checkpoint_audit.py \
pg_llm_batch/__init__.py \
tests \
docs \
CHANGELOG.md \
.github/workflows/ci.yml
printf '%s\n' '=== current implementation and API call sites ==='
rg -n -C 5 \
'build_audit_snapshot_manifest|_validate|transaction|autocommit|isolation|REPEATABLE|SERIALIZABLE|read.?only|digest|sha256|recorded_at|timestamp|offset|page_size|max_events|cursor|revalid' \
pg_llm_batch/checkpoint_audit.py pg_llm_batch/__init__.py tests docs CHANGELOG.md .github/workflows/ci.yml || true
printf '%s\n' '=== GitHub current PR metadata, reviews, and checks ==='
gh pr view 97 --repo ContextualWisdomLab/pg-llm-batch \
--json number,state,isDraft,headRefOid,baseRefOid,mergeStateStatus,reviews,statusCheckRollup
printf '%s\n' '=== GitHub unresolved review threads ==='
gh api graphql -F owner='ContextualWisdomLab' -F name='pg-llm-batch' -F number=97 -f query='
query($owner: String!, $name: String!, $number: Int!) {
repository(owner: $owner, name: $name) {
pullRequest(number: $number) {
reviewThreads(first: 100) {
nodes { isResolved comments(first: 1) { nodes { author { login } body path line } } }
}
}
}
}' \
--jq '[.data.repository.pullRequest.reviewThreads.nodes[] | select(.isResolved == false)]'Length of output: 202480 🏁 Script executed: #!/bin/bash
set -euo pipefail
base='f9062821556715537a028095326544bf234e45ae'
head='deed55787f4b7eb1ef57147071957f03631a3b1c'
printf '%s\n' '=== relevant symbols and line ranges ==='
ast-grep outline pg_llm_batch/checkpoint_audit.py --items all --view expanded \
| rg -n -C 2 \
'CheckpointAuditEvent|CheckpointAuditPage|CheckpointAuditSnapshotManifest|build_audit_snapshot_manifest|list_audit_event_page|_frame|_digest|_timestamp|transaction'
printf '%s\n' '=== head: manifest implementation region ==='
rg -n \
'^(def |class )|build_audit_snapshot_manifest|list_audit_event_page_in_transaction|_snapshot|_digest|_frame|_timestamp|transaction_status' \
pg_llm_batch/checkpoint_audit.py
sed -n '1,260p' pg_llm_batch/checkpoint_audit.py
sed -n '400,820p' pg_llm_batch/checkpoint_audit.py
printf '%s\n' '=== predecessor versus head page implementation ==='
git show "${base}:pg_llm_batch/checkpoint_audit.py" \
| rg -n -C 35 'def list_audit_event_page_in_transaction'
printf '%s\n' '--- head ---'
rg -n -C 35 'def list_audit_event_page_in_transaction' pg_llm_batch/checkpoint_audit.py
printf '%s\n' '=== focused tests: manifest, digest, integration ==='
for file in \
tests/test_checkpoint_audit_snapshot_manifest.py \
tests/test_checkpoint_audit_snapshot_digest_vector.py \
tests/test_checkpoint_audit_snapshot_transaction_boundary.py \
tests/test_checkpoint_audit_snapshot_integration.py \
tests/test_checkpoint_audit_snapshot_public_api.py
do
printf '\n=== %s symbols ===\n' "$file"
ast-grep outline "$file" --items all --view expanded
printf '\n=== %s focused assertions ===\n' "$file"
rg -n -C 12 \
'^def test_.*(overflow|page|digest|timestamp|offset|concurr|revalid|transaction|autocommit|serializ|repeatable)|assert .*snapshot|pytest\.raises' \
"$file"
done
printf '%s\n' '=== exact changes to pre-existing page method ==='
git diff --unified=30 "$base" "$head" -- pg_llm_batch/checkpoint_audit.py \
| sed -n '/list_audit_event_page_in_transaction/,+180p'Length of output: 57732 🏁 Script executed: #!/bin/bash
set -euo pipefail
printf '%s\n' '=== digest, timestamp, and transaction helper implementation ==='
nl -ba pg_llm_batch/checkpoint_audit.py | sed -n '250,360p'
printf '%s\n' '=== fixed digest compatibility vector ==='
nl -ba tests/test_checkpoint_audit_snapshot_digest_vector.py | sed -n '45,150p'
printf '%s\n' '=== live transaction and concurrent-insert sequence ==='
nl -ba tests/test_checkpoint_audit_snapshot_integration.py | sed -n '40,225p'
printf '%s\n' '=== timestamp offset and event field digest tests ==='
rg -n -C 20 -i \
'offset|timezone|recorded_at|timestamp|each.*field|field.*digest|digest.*field' \
tests/test_checkpoint_audit_snapshot_manifest.py \
tests/test_checkpoint_audit_snapshot_digest_vector.py \
tests/test_checkpoint_audit_snapshot_integration.py \
tests
printf '%s\n' '=== precise source range against predecessor ==='
git diff --unified=20 \
f9062821556715537a028095326544bf234e45ae \
deed55787f4b7eb1ef57147071957f03631a3b1c \
-- pg_llm_batch/checkpoint_audit.py pg_llm_batch/__init__.py \
CHANGELOG.md docs/checkpoint-audit.md docs/adr/0012-checkpoint-audit-snapshot-manifests.md \
docs/doctoring/checkpoint-audit-snapshot-manifests.md \
| sed -n '1,2200p'Length of output: 1638 🏁 Script executed: #!/bin/bash
set -euo pipefail
printf '%s\n' '=== digest, timestamp, and transaction helper implementation ==='
cat -n pg_llm_batch/checkpoint_audit.py | sed -n '250,360p'
printf '%s\n' '=== fixed digest compatibility vector ==='
cat -n tests/test_checkpoint_audit_snapshot_digest_vector.py | sed -n '45,150p'
printf '%s\n' '=== live transaction and concurrent-insert sequence ==='
cat -n tests/test_checkpoint_audit_snapshot_integration.py | sed -n '40,225p'
printf '%s\n' '=== timestamp and digest focused tests ==='
rg -n -C 16 -i \
'offset|timezone|recorded_at|timestamp|digest.*(field|chang)|field.*digest' \
tests/test_checkpoint_audit_snapshot_manifest.py \
tests/test_checkpoint_audit_snapshot_digest_vector.py \
tests/test_checkpoint_audit_snapshot_integration.py || true
printf '%s\n' '=== exact source and public API changes ==='
git diff --unified=20 \
f9062821556715537a028095326544bf234e45ae \
deed55787f4b7eb1ef57147071957f03631a3b1c \
-- pg_llm_batch/checkpoint_audit.py pg_llm_batch/__init__.py \
| sed -n '1,1800p'Length of output: 59875
The manifest uses bounded keyset pages and fixed digest state. It fails closed when more events remain after The SHA-256 framing is length-prefixed and domain-separated. It binds the trusted key, every retained event field, normalized UTC microsecond timestamps, final event count, and identity bounds. It excludes page partitioning. The fixed schema-version-1 digest vector protects this framing contract. The timestamp guard rejects The live PostgreSQL test establishes the stable read-only snapshot before another connection commits an event. It verifies page-size invariance, later-event exclusion, The package-root exports and the reviewed documentation state the public API and transaction boundary. GitHub currently reports that This is staged Draft source and current-stack evidence only. It does not transfer evidence from closed superseded
|
|
SUPERSEDED by current-main reconstruction #181 together with its obsolete predecessor line. Fresh evidence shows #97 still depends on #96, while #96 depends on closed-unmerged #95 and #92/#94/#95 are closed unmerged. Preserve this branch only as design/test provenance; do not retarget it or transfer its historical CI/review evidence onto the current protected-main reconstruction. |
1 participant
Linearized replacement for stale #84
This Draft rebuilds the checkpoint-audit snapshot-manifest feature on exact current pagination predecessor #96 rather than extending stale #83 ancestry. #84 is now closed unmerged as SUPERSEDED; none of its checks, reviews, or approvals transfer.
RCA and replacement boundary
The branch starts from #96 exact head
f9062821556715537a028095326544bf234e45ae. Before replay, the old #83 and current #96 baselines were verified byte-identical for the unaffected snapshot contract surfaces.pg_llm_batch/checkpoint_audit.pyrequired deliberate composition because the current stack contains exact requested-key revalidation that #84 did not, andCHANGELOG.mdrequired composition because the current stack contains newer accepted entries.The replacement preserves #96 pagination/key-revalidation behavior and adds #84's bounded snapshot-manifest API, digest framing, transaction/isolation checks, timestamp-offset hardening, public exports, CI/live PostgreSQL regressions, ADR/operator/doctoring contracts, and final authoritative CHANGELOG wording. The caller must supply one active PostgreSQL transaction that is read-only and uses
REPEATABLE READorSERIALIZABLE; autocommit and session-level-only isolation are rejected.Test-first evidence
The replacement established the snapshot contract on current #96 ancestry before production composition. After implementation, exact-head CI exposed one remaining documentation contract failure: CHANGELOG had not stated the active-transaction/autocommit/isolation boundary. Commit
deed55787f4b7eb1ef57147071957f03631a3b1ccorrected that authoritative contract without changing production behavior.Current exact staged evidence
deed55787f4b7eb1ef57147071957f03631a3b1c.f9062821556715537a028095326544bf234e45ae.31319990050: completed / success.31319990023: completed / success.These are staged replacement checks, not protected-main acceptance. Final security/default-branch/review/policy evidence still depends on the predecessor stack reaching protected main and this unchanged final source being revalidated against the integrated base.
Dependency and merge boundary
Required order remains
.github#790 -> pg-llm-batch#53 -> #57 -> #58 -> #59 -> #60 -> #92 -> #94 -> #95 -> #96 -> this replacement.Keep Draft while predecessors remain unintegrated. After prerequisite integration, reconcile/retarget only when integration correctness requires it and regenerate every applicable exact-head/exact-base CI, live PostgreSQL transaction/concurrency, security, dependency, SBOM/provenance, packaging, automated-review, branch-protection, repository-policy, and Release Acceptance gate. Merge only with zero valid unresolved findings and a qualifying independent non-author formal
APPROVEDreview on the unchanged final head where required. Queued, pending, cancelled, skipped-required, absent, neutral-required, stale-head/base, predecessor, synthetic-only, status-only, author-only, rate-limited, or failed evidence is not acceptance.