Evidence

What the audited run actually covered

Every claim on this site sits at one of five evidence levels, named where the claim is made. The source directory kept changing during development, so the tests ran against a frozen copy with a per-file hash manifest — 226 files, captured at 20:12:15 UTC on 18 September 2026. The earlier 19:32 snapshot is not retroactively treated as if it already contained the later code.

LevelWhat it supportsWhat it does not support
Source presentA function and its contract exist in the frozen versionThat every path is correct, or that it runs in production
Functionally testedConcrete fixtures, error cases and assertions pass on this hostUniversal correctness, proven performance, or recovery from every hardware fault
Historically measuredA stored run holds results for one fixed workload and measurement boundaryThe same outcome on another machine, source version or dataset
Hash checkedThe bytes checked equal the bytes named in a receiptIndependent authentication of the author, or a correct measurement method
Still openThe evidence was not found here, or was never producedThat the thing never happens anywhere outside this repository
$ CARGO_TARGET_DIR=/…/gqodb/target \
  cargo test --workspace --locked --offline --lib --tests -j 2

# 282 passed · 0 failed · 0 ignored · 85 completed test targets

This does not re-accept doc-tests, every example, Clippy, formatting or a release build. Workspace tests do include the short, explicitly unqualified IPC selftest.

The kill test, in detail

engine test · process-kill recovery
What it does

Imports 8,192 records, waits for a partially committed durable prefix, kills its own subprocess, reopens, recovers, and completes the import.

What it asserts

Record count, a single input charge, the recovered raw SHA-256, and recognition of the re-import. Strong evidence for this import path under process failure.

What it is not

Not a physical power-loss test, and not a broker reconnecting mid-stream. The test environment is temporary local storage.

Hash checks

30 result files matched the hashes in their historical build receipts. That proves byte agreement with a chosen digest — not that the measurement method was right.

What else passed

cargo-test.log
✓Raw preservation on invalid data, missing clock evidence, exclusive writer and config binding.
✓Refusing wholly corrupt frames while distinguishing a physically incomplete tail.
✓Refused quota and buffer limits, a failed ledger publication, and an interrupted receipt delivery.
✓Background derivation, worker restart, and raw capture continuing through a visible compression failure.
✓Exact native times and metadata, large unsigned values, sliced rows, and codec profile selection from the header.
✓Causal queries, late corrections, tombstones, clock intervals and dependencies.
✓Exact Arrow and Parquet exchange, model and news round-trips, and fixture replay.
✓Paper-order idempotency, cancel races, and separation of external constraints from internal targets.
✓Immutable experiment manifests and detection of changed inputs or results.
✓Ops: missing observations, clock staleness, counter resets, worker stalls and exact manual deadlines.
✓Installer: temporary fixture install, relocatability, hash and path checks, rollback and targeted uninstall.

An installer fixture is not a complete published suite installation, and a kill test of a bounded import is not a live-feed soak test. Green tests do not widen their own scope.

What this audit did not do

No live feed, no trading service, no training run, no long latency qualification, and no physical host-clock calibration were started. It is new functional test evidence plus a recomputation of existing measurements — not a new performance benchmark and not a production certification.

Evidence fileContents
source-manifest.jsonThe frozen source copy: 226 files with per-file hashes
cargo-test.logThe full test log behind the 282 / 0 / 0 result
test-result.jsonMachine-readable test summary
verified-metrics.jsonRecomputed benchmark figures and the receipt hash checks
verify_evidence.pyThe reproducible check script

All in the private engine repository.