feat(parquet): audit-stream writer/reader (PR-G) - #46
Merged
Conversation
Lands the audit half of RFC 0005 §3.10's crate-shape plan. Closes
RFC0005.7 (audit-event stream is a separate file series, every §3.7
row-level column round-trips) and the audit-axis half of RFC0005.11
(row-vs-path validation on partition mismatch).
- `AuditWriter::open` / `append_events` / `close` mirroring the data
writer's atomic publish (`.parquet.tmp` + rename-on-close + `Drop`
cleanup). Writes to `partition.audit_path()`, which stops at
`day=DD/` per §3.4 — one axis coarser than the data partitioning.
- `AuditReader::open_partition` / `open_file` / `read_all` enforcing
the §3.9 schema-evolution contract on the audit side. Row-vs-path
validation compares only tenant + year/month/day; the hour field
is populated on the shared `PartitionKey` shape but ignored on the
audit axis.
- `audit_events_to_batch` building the §3.7 columns from a slice of
`AuditEvent`. Per-variant column population follows the §3.7
mapping table:
* `TemplateWidened` → ordinal 0, positions populated, slots empty
* `TemplateTypeExpanded` → ordinal 1, positions empty, slots populated
* `TemplateWideningRejectedDegenerate` → ordinal 2, both empty,
`would_be_template` + `would_be_positions` JSON-encoded into the
`reason` column (§3.7's "diagnostic string" framing accommodates
the structured payload without amending the schema).
- §3.7 encoding policy: ZSTD-3 codec, dict-off on `timestamp` /
`old_template` / `new_template` / `triggering_line_hash` /
`triggering_line_sample`, page index downgraded to chunk-only on
every column except the §3.7 `Page index = yes` four
(`timestamp`, `event_kind`, `event_type`, `template_id`).
Adds `serde_json` as a dep for the rejection-variant reason payload.
Tests:
- `tests/audit_round_trip.rs` — RFC0005.7 full round-trip of one of
each variant; sub-tests pinning the audit-path-stops-at-day layout
and the rejection-variant `would_be_*` survival via `reason`.
- `tests/audit_row_vs_path_validation.rs` — RFC0005.11 audit-side
mirror: tenant mismatch, day mismatch, same-day-different-hour
validates cleanly, `open_file` skips validation.
Known gap: unknown `event_kind` ordinals surface as a hard
`AuditReaderError::UnknownEventKind` rather than an
`AuditEventKind::Unknown` variant (no analog of `ParamType::Unknown`
on the audit-event enum yet). Deferred until a real new variant
lands via a §3.8 amendment.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughAdds audit-stream Parquet I/O per RFC 0005: event-to-batch serialization (JSON-encoded rejection diagnostics), an atomic-file Parquet writer with partition validation, a reader that decodes rows and enforces per-row partition checks, and integration tests validating round-trip and partition behavior. ChangesAudit Stream Parquet I/O
Sequence Diagram(s)sequenceDiagram
participant Client
participant AuditWriter
participant ArrowWriter
participant Filesystem
Client->>AuditWriter: open(partition)
Client->>AuditWriter: append_events(events)
AuditWriter->>ArrowWriter: write RecordBatch
ArrowWriter->>Filesystem: emit .parquet.tmp
Client->>AuditWriter: close()
AuditWriter->>Filesystem: rename .parquet.tmp -> .parquet
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Contributor
There was a problem hiding this comment.
Pull request overview
Implements the audit-event Parquet file series described in RFC 0005 by adding a dedicated audit writer/reader pair plus an Arrow RecordBatch builder and round-trip/validation tests, completing the Parquet crate’s audit-stream functionality.
Changes:
- Added
AuditWriterto write audit-event Parquet files under the day-partitionedaudit/…/day=DD/path with atomic publish-on-close semantics. - Added
AuditReaderto read audit-event Parquet files with RFC §3.9 schema-evolution behavior and audit-axis row-vs-path validation (tenant/year/month/day; hour ignored). - Added audit-stream batch construction (
audit_events_to_batch) and end-to-end tests, plusserde_jsonfor encoding/decoding the rejection variant’sreasonpayload.
Reviewed changes
Copilot reviewed 7 out of 8 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| crates/ourios-parquet/src/lib.rs | Exposes new audit modules and re-exports audit reader/writer/batch APIs. |
| crates/ourios-parquet/src/audit_writer.rs | Implements audit Parquet writer with RFC §3.7 encoding policy and audit-axis partition validation. |
| crates/ourios-parquet/src/audit_reader.rs | Implements audit Parquet reader with schema evolution rules and row-vs-path validation. |
| crates/ourios-parquet/src/audit_record_batch.rs | Builds audit Arrow RecordBatch rows per RFC §3.7 mapping, including rejection reason JSON encoding. |
| crates/ourios-parquet/tests/audit_round_trip.rs | Adds end-to-end audit writer→reader round-trip tests and audit path layout assertions. |
| crates/ourios-parquet/tests/audit_row_vs_path_validation.rs | Adds row-vs-path validation tests for tenant/day mismatch and same-day different-hour behavior. |
| crates/ourios-parquet/Cargo.toml | Adds serde_json dependency to support rejection-variant reason encoding/decoding. |
| Cargo.lock | Updates lockfile for the new dependency. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
🧹 Nitpick comments (1)
crates/ourios-parquet/src/audit_writer.rs (1)
319-347: 💤 Low valueConsider adding colocated unit tests for partition helpers.
derive_audit_partitionandaudit_partition_matchescontain non-trivial logic (timestamp conversion, partition comparison ignoring hour). While integration tests cover the end-to-end flow, colocated unit tests would provide faster feedback and clearer coverage of edge cases like DST boundaries, leap years, or tenant ID comparisons.As per coding guidelines: "Unit tests must be colocated next to the code for anything non-trivial."
🤖 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/ourios-parquet/src/audit_writer.rs` around lines 319 - 347, Add colocated unit tests next to derive_audit_partition and audit_partition_matches that exercise timestamp conversion and comparison logic: create AuditEvent instances to assert derive_audit_partition returns expected PartitionKey for normal timestamps (including leap day and DST-transition times), asserts handling of pre-epoch timestamps produces AuditWriterError::Batch(AuditBatchError::PreEpochTimestamp), tests timestamp overflow path producing AuditWriterError::Batch(AuditBatchError::TimestampOverflow), and verify audit_partition_matches ignores hour by asserting equality for same tenant/year/month/day with different hours and inequality for differing tenant, year, month, or day; reference the functions derive_audit_partition, audit_partition_matches and the error types AuditWriterError/AuditBatchError when adding these unit tests.
🤖 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.
Nitpick comments:
In `@crates/ourios-parquet/src/audit_writer.rs`:
- Around line 319-347: Add colocated unit tests next to derive_audit_partition
and audit_partition_matches that exercise timestamp conversion and comparison
logic: create AuditEvent instances to assert derive_audit_partition returns
expected PartitionKey for normal timestamps (including leap day and
DST-transition times), asserts handling of pre-epoch timestamps produces
AuditWriterError::Batch(AuditBatchError::PreEpochTimestamp), tests timestamp
overflow path producing
AuditWriterError::Batch(AuditBatchError::TimestampOverflow), and verify
audit_partition_matches ignores hour by asserting equality for same
tenant/year/month/day with different hours and inequality for differing tenant,
year, month, or day; reference the functions derive_audit_partition,
audit_partition_matches and the error types AuditWriterError/AuditBatchError
when adding these unit tests.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 910d68b5-6d44-405a-a486-a49623f6fe44
⛔ Files ignored due to path filters (1)
Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (7)
crates/ourios-parquet/Cargo.tomlcrates/ourios-parquet/src/audit_reader.rscrates/ourios-parquet/src/audit_record_batch.rscrates/ourios-parquet/src/audit_writer.rscrates/ourios-parquet/src/lib.rscrates/ourios-parquet/tests/audit_round_trip.rscrates/ourios-parquet/tests/audit_row_vs_path_validation.rs
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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/ourios-parquet/src/audit_reader.rs`:
- Line 128: The batch-to-file-global row indexing change is incomplete: call
sites that still produce batch-local "row {i}" must receive and use the
file-global offset. Update batch_to_audit_events usage so it passes row_offset
into the downstream helpers (or change the helpers to accept row_offset):
required_string, required_timestamp, required_fixed_bytes16, and
materialize_required_primitive should accept a row_offset parameter (or a single
normalized function that computes file_row = row_offset + i) and use file_row in
Conversion/error messages; propagate that parameter through the call chain for
the other affected spots (around the blocks you changed at batch_to_audit_events
and the ranges referenced) so all Conversion errors report file-global rows
consistently.
- Around line 288-297: Add colocated unit tests in the same file that directly
exercise batch_to_audit_events: (1) construct a multi-batch scenario (multiple
RecordBatch inputs with a non-first-row error) and assert the returned
AuditReaderError contains a file-global row index matching row_offset +
local_row_index (verifying the multi-batch error-index path and
PartitionMismatch behavior), and (2) craft a RecordBatch with a timestamp value
that triggers overflow in the checked_add path and assert the function returns
the specific overflow AuditReaderError (the checked_add failure). Use the
existing AuditEvent, AuditReaderError, RecordBatch types and the
batch_to_audit_events helper to drive these tests and place them adjacent to the
function (also cover the analogous cases noted around the other related
functions at 379-389).
🪄 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: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: c0883e86-5d34-449f-99b6-94193309eb48
⛔ Files ignored due to path filters (1)
Cargo.lockis excluded by!**/*.lock
📒 Files selected for processing (3)
crates/ourios-parquet/Cargo.tomlcrates/ourios-parquet/src/audit_reader.rscrates/ourios-parquet/src/audit_record_batch.rs
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
crates/ourios-parquet/src/audit_writer.rs (1)
80-261: 🛠️ Refactor suggestion | 🟠 Major | ⚡ Quick winAdd colocated unit coverage for this module’s invariants.
This is a non-trivial writer module, but there are no inline tests here. Please add
#[cfg(test)]coverage next to the code for the audit partition matching rules and the temp-file lifecycle on close/error paths.As per coding guidelines, "Unit tests must be colocated next to the code for anything non-trivial".
Also applies to: 357-464
🤖 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/ourios-parquet/src/audit_writer.rs` around lines 80 - 261, Add colocated #[cfg(test)] unit tests in this module that (1) exercise the partition-matching invariants by calling derive_audit_partition on synthesized AuditEvent structs and asserting audit_partition_matches against PartitionKey instances used to open an AuditWriter (use AuditWriter::open to get a real partition path where appropriate) and (2) cover temp-file lifecycle by creating an AuditWriter (AuditWriter::open), writing some events with append_events, then forcing both a successful close (AuditWriter::close) to assert the .parquet.tmp is renamed to final_path and a failure path (e.g., mock or induce ArrowWriter::try_new/write/close to fail or simulate an IO rename error) to assert the .parquet.tmp remains on disk after the error; reference the symbols AuditWriter::open, append_events, close, audit_partition_matches, derive_audit_partition, final_path and temp_path in the tests so they exercise the actual module invariants.
🤖 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/ourios-parquet/src/audit_writer.rs`:
- Around line 184-195: The loop handling Parquet writes must mark the writer as
poisoned when inner.flush() or inner.write() fails so close() won't publish a
partial file; modify the implementation (e.g., in the method containing the
shown loop, likely append_events) to set a boolean flag on the AuditWriter
instance (name it poisoned or similar) before returning an Err when
inner.flush() or inner.write() returns an error, and update AuditWriter::close
to check that poisoned flag and refuse to rename/publish the temp file
(returning an error) if set; ensure the error returned from the failing Parquet
call is preserved when setting the flag and propagated.
---
Outside diff comments:
In `@crates/ourios-parquet/src/audit_writer.rs`:
- Around line 80-261: Add colocated #[cfg(test)] unit tests in this module that
(1) exercise the partition-matching invariants by calling derive_audit_partition
on synthesized AuditEvent structs and asserting audit_partition_matches against
PartitionKey instances used to open an AuditWriter (use AuditWriter::open to get
a real partition path where appropriate) and (2) cover temp-file lifecycle by
creating an AuditWriter (AuditWriter::open), writing some events with
append_events, then forcing both a successful close (AuditWriter::close) to
assert the .parquet.tmp is renamed to final_path and a failure path (e.g., mock
or induce ArrowWriter::try_new/write/close to fail or simulate an IO rename
error) to assert the .parquet.tmp remains on disk after the error; reference the
symbols AuditWriter::open, append_events, close, audit_partition_matches,
derive_audit_partition, final_path and temp_path in the tests so they exercise
the actual module invariants.
🪄 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: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 302b8ad1-74e4-4d0f-a5d7-c990df34350c
📒 Files selected for processing (1)
crates/ourios-parquet/src/audit_writer.rs
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
🧹 Nitpick comments (1)
crates/ourios-parquet/src/audit_record_batch.rs (1)
498-536: 💤 Low valueConsider adding a unit test for the
TemplateMustNotChangeerror path.The new invariant check for
TemplateTypeExpanded(lines 299-305) is non-trivial validation logic. A colocated test that constructs aTemplateTypeExpandedevent with divergentold_template/new_templateand asserts the expectedAuditBatchError::TemplateMustNotChangeerror would pin this behavior.💡 Suggested test
#[test] fn template_type_expanded_rejects_divergent_templates() { let e = AuditEvent { kind: AuditEventKind::TemplateTypeExpanded { old_version: 2, new_version: 3, old_template: "[\"user\",\"<*>\"]".to_string(), new_template: "[\"different\",\"<*>\"]".to_string(), // divergent slots_expanded: vec![], }, tenant_id: TenantId::new("acme"), template_id: 7, triggering_line_hash: hash_triggering_line(b"trigger"), triggering_line_sample: None, timestamp: ts(1_775_127_480), }; let err = audit_events_to_batch(std::slice::from_ref(&e)) .expect_err("divergent templates must error"); assert!(matches!(err, AuditBatchError::TemplateMustNotChange { .. })); }As per coding guidelines,
crates/**/src/**/*.rs: Unit tests must be colocated next to the code for anything non-trivial.🤖 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/ourios-parquet/src/audit_record_batch.rs` around lines 498 - 536, Add a colocated unit test that constructs an AuditEvent with kind AuditEventKind::TemplateTypeExpanded where old_template and new_template are different, calls audit_events_to_batch(std::slice::from_ref(&e)), and asserts it returns an Err matching AuditBatchError::TemplateMustNotChange; specifically reference the TemplateTypeExpanded variant, the audit_events_to_batch function, and the AuditBatchError::TemplateMustNotChange pattern to locate the code to test and ensure the divergent templates trigger the expected error.
🤖 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.
Nitpick comments:
In `@crates/ourios-parquet/src/audit_record_batch.rs`:
- Around line 498-536: Add a colocated unit test that constructs an AuditEvent
with kind AuditEventKind::TemplateTypeExpanded where old_template and
new_template are different, calls
audit_events_to_batch(std::slice::from_ref(&e)), and asserts it returns an Err
matching AuditBatchError::TemplateMustNotChange; specifically reference the
TemplateTypeExpanded variant, the audit_events_to_batch function, and the
AuditBatchError::TemplateMustNotChange pattern to locate the code to test and
ensure the divergent templates trigger the expected error.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro Plus
Run ID: 2b1797d2-20ad-4319-976a-dbdc575b7319
📒 Files selected for processing (4)
crates/ourios-parquet/src/audit_reader.rscrates/ourios-parquet/src/audit_record_batch.rscrates/ourios-parquet/src/audit_writer.rscrates/ourios-parquet/src/writer.rs
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 8 out of 9 changed files in this pull request and generated 2 comments.
Comments suppressed due to low confidence (2)
crates/ourios-parquet/src/writer.rs:254
Writer::append_recordscan still be called after a prior Parquet error has setself.poisoned = true. Since the ArrowWriter buffer is documented as undefined after a failedwrite/flush, subsequent appends should refuse immediately (e.g., returnWriterError::Poisoned) rather than attempting more writes into a potentially-corrupted in-memory state.
pub fn append_records(&mut self, records: &[MinedRecord]) -> Result<(), WriterError> {
if records.is_empty() {
return Ok(());
}
for (idx, r) in records.iter().enumerate() {
let derived = PartitionKey::derive(r).map_err(|e| WriterError::Batch(e.into()))?;
if derived != self.partition {
return Err(WriterError::PartitionMismatch {
row_index: idx,
expected: self.partition.clone(),
actual: derived,
});
}
}
let inner = self
.inner
.as_mut()
.expect("inner ArrowWriter is Some until Writer::close is called");
// Run the Parquet-touching loop in a helper that takes a
// `&mut ArrowWriter<File>` so the outer `self.poisoned =
// true` assignment can run after the borrow on `self.inner`
// ends. Poison only on Parquet errors — `Batch` errors
// come from `mined_records_to_batch` BEFORE any
// `inner.write` touches the buffer, so the inner writer
// is still in a clean state and a follow-up
// `append_records` is safe.
let result = append_chunks(inner, records);
if matches!(result, Err(WriterError::Parquet(_))) {
self.poisoned = true;
}
result
}
crates/ourios-parquet/src/audit_writer.rs:217
AuditWriter::append_eventsdoes not checkself.poisonedon entry, so callers can keep appending after a Parquetwrite/flusherror even though the ArrowWriter buffer is documented as undefined. Consider returningAuditWriterError::Poisonedimmediately whenpoisonedis set to prevent further writes into a potentially-corrupted state.
pub fn append_events(&mut self, events: &[AuditEvent]) -> Result<(), AuditWriterError> {
if events.is_empty() {
return Ok(());
}
for (idx, e) in events.iter().enumerate() {
let derived = derive_audit_partition(e)?;
if !audit_partition_matches(&derived, &self.partition) {
return Err(AuditWriterError::PartitionMismatch {
row_index: idx,
expected: self.partition.clone(),
actual: derived,
});
}
}
let inner = self
.inner
.as_mut()
.expect("inner ArrowWriter is Some until AuditWriter::close is called");
// Run the Parquet-touching loop in a helper that takes a
// `&mut ArrowWriter<File>` so the outer `self.poisoned =
// true` assignment can run after the borrow on `self.inner`
// ends. Poison only on Parquet errors — `Batch` errors
// come from `audit_events_to_batch` BEFORE any `inner.write`
// touches the buffer, so the inner writer is still in a
// clean state and a follow-up `append_events` is safe.
let result = append_chunks(inner, events);
if matches!(result, Err(AuditWriterError::Parquet(_))) {
self.poisoned = true;
}
result
}
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This was referenced Jun 4, 2026
This was referenced Jun 28, 2026
Merged
This was referenced Jul 6, 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
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.
Summary
Lands the audit half of RFC 0005 §3.10's crate-shape plan, closing out the Parquet crate's audit-event file series.
AuditWriter(production) — opens an audit Parquet file underpartition.audit_path()(one axis coarser than the data path: stops atday=DD/), appendsAuditEvents, atomically publishes on close. Same.parquet.tmp+ rename-on-close +Dropcleanup story as the data writer.AuditReader::open_partition/open_file/read_all— §3.9 schema-evolution contract on the audit side. Row-vs-path validation compares tenant + year/month/day (the hour field is ignored on the audit axis since the audit path has no hour segment).audit_events_to_batch— builds the §3.7 column set per the normative mapping table. Per-variant column population:TemplateWidened→ ordinal 0, positions populated, slots emptyTemplateTypeExpanded→ ordinal 1, positions empty, slots populatedTemplateWideningRejectedDegenerate→ ordinal 2; the in-memory variant'swould_be_template/would_be_positionsride in the §3.7reasoncolumn as a JSON object. The §3.7 text framesreasonas "the degenerate-template guard's diagnostic string"; encoding it as JSON preserves the in-memory shape without amending the schema.timestamp/old_template/new_template/triggering_line_hash/triggering_line_sample, page-index downgraded to chunk on every column except the §3.7Page index = yesfour (timestamp,event_kind,event_type,template_id).serde_jsonas aourios-parquetdependency for the rejection-variantreasonpayload encoder/decoder.RFC criteria closed
Tests added
tests/audit_round_trip.rs— one event of each variant throughAuditWriter→AuditReader::open_partitionwith full struct equality; sub-tests pin the audit-path layout (nohour=segment) and the rejection variant'swould_be_*survival.tests/audit_row_vs_path_validation.rs— tenant mismatch / day mismatch surfacePartitionMismatch; same-day-different-hour validates cleanly;open_fileskips validation.#[cfg(test)] mod testsinsrc/audit_record_batch.rsfor the batch builder + the rejection-reason JSON round-trip + the pre-epoch timestamp rejection.Known gap
Unknown
event_kindordinals surface as a hardAuditReaderError::UnknownEventKindrather than anAuditEventKind::Unknownvariant — there is no analog ofParamType::Unknownon the audit-event enum yet. Deferred until a real new variant lands via a §3.8 amendment, which will either extendAuditEventKinddirectly or introduce theUnknown(u8)catch-all.Test plan
cargo fmt --all --checkcargo clippy --all-targets --all-features -- -D warningscargo test --all-features(32 + 116 + 9 + 7 + 4 + 7 + 26 + 3 + 4 + 2 + 4 + 2 + 4 + 2 = all green; RFC0005.7 + audit RFC0005.11 pass)ourios-ingester's audit sink →AuditWriter) is out of scope for this PR.🤖 Generated with Claude Code
Summary by CodeRabbit
New Features
Bug Fixes / Reliability
Tests