C28 phase 2 (Stratification); realizes item 1 of the deferred #147. The
engine's production surface no longer names a backtest-metric type:
- RunReport becomes generic over its metric payload M; sweep/mc/walkforward/
blueprint thread the parameter (SweepPoint<M>, SweepFamily<M>, WindowRun<M>,
WalkForwardResult<M>). RunManifest stays concrete and engine-owned (its
selection: Option<FamilySelection> embeds the foundation-grade analysis type).
- summarize and the MC assembly (McDraw/McFamily/McAggregate/RBootstrap/
r_bootstrap/monte_carlo) move to aura-backtest - McAggregate::from_draws reads
RunMetrics fields by name, so generifying it is the phase-6 metric-vocabulary
abstraction (#147 item 2), still deferred; wholesale relocation is the honest
cut. The concrete instantiation lives in aura-backtest as
`type RunReport = aura_engine::RunReport<RunMetrics>` + sibling aliases.
- the statistics kernel (MetricStats/quantile/resample_block/SplitMix64) moves
to the aura-analysis foundation; the engine re-imports it (inner->foundation,
legal) and re-exports it so existing consumers stay source-compatible.
Dependency inversion in one commit: aura-engine drops aura-backtest from
[dependencies] (back to dev-deps for its SimBroker/RunMetrics test fixtures);
aura-backtest gains aura-engine. Cycle-free for lib targets - the cycle closes
only through the engine's dev-dep edge, the pattern aura-vocabulary already
uses. aura-backtest reaches the kernel transitively through the engine
re-export, so no aura-backtest -> aura-analysis edge exists (the C28 ladder
permits backtest -> {core, engine} only). run_indexed / SplitMix64::next_f64
widened pub(crate) -> pub for cross-crate use.
Consumers (registry/campaign/cli/composites/ingest/bench) rewired by import
path only, no call-site logic changed. The c28_layering structural test extends
to the full ladder: aura-analysis (no aura-* deps), aura-engine ⊆ {core,
analysis}, aura-backtest ⊆ {core, engine}.
Behaviour-preserving: 1448/0 tests, clippy -D warnings clean, serde shapes
byte-identical (C18 - RunReport<M> keeps field order manifest,metrics; the
CLI pre-serialized-splice contract unchanged), moved code traceable via git
rename detection. Cycle-introduced broken intra-doc links fixed.
closes#292
closes#272
A member fault (no-data, bind, run, or a caught panic) is now a recorded
per-cell outcome instead of aborting the whole campaign and discarding every
already-computed cell. The incident that motivated this (a 22-instrument
campaign lost ~36 healthy cells ~6.7 min in because Copper had an archive gap)
now completes: the healthy cells persist, the gap cell is recorded as failed,
and the run exits 3.
Direction (owner decision 2026-07-14): run to completion and report
compromised results; no coverage preflight, no window synthesis.
Containment granularity:
- The CELL for a sweep-stage member fault (a grid hole structurally
compromises winner selection, so the whole cell fails).
- The FOLD for a walk_forward member fault (independent time windows): the
surviving folds pool into the family, failed folds are recorded as
StageRealization.window_faults, and the summary names the ratio.
- aura-registry: additive CellFault / CellFaultKind (closed:
no_data|bind|run|panic|window) / WindowFault / CellCoverage, plus
fault/coverage fields on CellRealization and window_faults on
StageRealization — all serde-default-skipped, so pre-#272 campaign_runs
lines parse and round-trip byte-identical.
- aura-campaign: run_cell returns a fault-annotated CellRealization instead of
Err (execute's accumulate-then-append-once tail is unchanged and now
persists every healthy cell + the one run record); a `contain` split keeps
ExecFault::Registry and doc-shape preflight faults global while Member/Window
become per-cell/per-fold. Member panics are caught with
catch_unwind(AssertUnwindSafe) at all three member-run sites (sweep IS/OOS)
and recorded as MemberFault::Panic — a member panic no longer aborts the
process. The wf stage partitions Registry faults (global) from Member/Window
(per-fold) and filters faulted-fold placeholders (the
"faulted-member-placeholder" broker sentinel) out of the persisted family.
- aura-cli: exec_fault_prose gains the Panic arm; CliMemberRunner::window_coverage
derives effective bounds + interior gap months from the #264 archive
primitives; present_campaign prints per-cell failure notes + a completion
summary and threads the failed-cell count; a run with >=1 failed cell exits 3
("completed with failed cells") uniformly across `aura campaign run` and the
dissolved sweep/walkforward/mc/generalize verbs (exit_on_campaign_result).
Usage stays 2, refused-before-running stays 1, clean stays 0.
Tests: the global-abort pins flip to containment (execute + the two wf fault
tests → fold-containment + all-folds-fail-the-cell); new panic-containment
tests on both the sweep path (PanicRunner) and the wf path (this commit adds
the wf mirror the loop left uncovered); a new gapped-archive e2e (one covered
cell + one gap cell → exit 3); the ~14 CLI exit-1 pins move to the exit-3
register; a pre-#272-line byte-identical round-trip guard.
Suite: cargo test --workspace green (1309 tests, 0 failed); clippy clean.
Decision log: #272 comments (fork rationale, the fold Registry/Member split,
the placeholder sentinel, uniform exit-3).
Follow-up (minor, not blocking): the plan under-scoped Task 1 to aura-registry
though the additive fields also touch aura-campaign's exec.rs literals — the
loop absorbed it mechanically; a future plan for a cross-crate additive-field
change should scope every crate's construction sites in the first task.
Every hand-rolled test-sandbox helper keyed its directory on
std::process::id() under env::temp_dir(), so the pre-create wipe never
matched a prior run's path: one directory stranded per helper site per
test-binary invocation, >50k dirs / 15.5 GiB on the 16 GiB tmpfs by
2026-07-13.
The remedy extends commit 84e1075's knob-lab pattern to every site:
fixed, tag-keyed, PID-FREE names under the build-tree tmp anchor, each
site keeping its pre-create remove_dir_all — which now genuinely
reclaims the previous run. Integration/bench targets use
env!("CARGO_TARGET_TMPDIR"); src-internal unit-test helpers (cargo sets
that var only for integration/bench targets) derive the same anchor
from env!("CARGO_MANIFEST_DIR") + ../../target/tmp. Residue is bounded
to one directory per site, on disk instead of the tmpfs, per checkout.
One deliberate exception: project_new.rs's tmp() stays under
env::temp_dir() because two of its tests (new_outside_a_work_tree_*)
need a base genuinely detached from any git work tree, and target/
lives inside the checkout's work tree. Its name carries a per-checkout
hash discriminator instead, so concurrently running checkouts (which
share /tmp) cannot race on the fixed name while each run still
reclaims its predecessor.
tempfile-RAII was considered and rejected (issue #258 decision log):
it cannot cover the process-lifetime OnceLock scaffolds (statics never
drop), adds a dependency, and still strands kill-leftovers on the
tmpfs.
Verified: full workspace suite green (incl. the new RED regression
test temp_cwd_sandbox_does_not_leak_under_env_temp_dir), clippy clean,
sandboxes observed under target/tmp with stable names across runs.
Stale pre-fix PID-keyed dirs in /tmp are not swept by this change and
age out only by manual cleanup.
closes#258
RunManifest gains defaults: Vec<(String, Scalar)> — the wrap-prefixed
bound_param_space() of the signal, read after axis reopening, so a
bound param an axis overrode has already left the space and flows
through params instead (disjoint by construction; verified end to end:
a sweep member's overridden fast.length sits in params while
slow.length/bias.scale sit in defaults). params keeps its "what varied"
semantics and stays the reproduce input. One-directional serde
widening (#[serde(default)]) per the selection/instrument/topology_hash
idiom — old records deserialize with an empty defaults; unlike the
Option fields it always serializes, mirroring params. ~20
struct-literal sites across five crates gained the field
(compile-mandated breadth, no behaviour change at those sites).
The C14 ledger records the underlying decision (2026-07-13): generated
outcome records spend redundancy on direct readability (single writer,
cannot drift); authored intent artifacts admit none (every redundancy
is a drift site) — so the fix lands in the manifest, never the
blueprint.
Verification: RED test run_manifest_stamps_untouched_bound_defaults
green; cargo build --workspace; cargo test --workspace green; clippy
-D warnings on the touched crates; binary-level sweep exclusivity
check.
The persist-side C1 drift alarm re-ran every nominee under the hardcoded
default vol-stop because CellRealization dropped the regime at the record
boundary — a non-default risk regime + persist taps false-failed C1 and
refused to persist a correct trace; two regimes over the same cell would
also have collided in one trace dir.
Fix at the record boundary: CellRealization carries regime + regime_ordinal
(serde-default + skip-serializing keeps pre-existing stored records
byte-stable), populated where the cell is realized; the regime->StopRule
binding is single-sourced in stop_rule_for_regime, shared by the member
runner and the persist re-run so the two sites cannot drift apart again;
campaign_cell_key appends -r{regime_ordinal} for non-default regimes
(default keys unchanged, no stored dir renamed).
closes#219, closes#212
Task 1: the tap namespace is a closed vocabulary — tap_vocabulary()
(the wrap convention's four sink names; escalation = new entry or an
authored blueprint sink, never an open node-path namespace) with
DocFault::UnknownTap in validate_campaign (index-addressed, prose
enumerating the vocabulary via the single source), SlotKind::TapKinds
on the persist_taps slot, open-slot hint naming the taps. The 0107
deferral seam test split into its two 0109 pins (unknown tap refuses
at validate; a valid tap passes validate and hits the member-data
seam) — the deferral eprintln itself survives until task 3.
Task 2: CampaignRunRecord.trace_name lands serde-default sparse; the
claim contract holds across the seam — execute stamps the sentinel iff
persist_taps is non-empty, append_campaign_run composes
Some("{campaign8}-{run}") onto the stored line (safe prefix take at
the registry seam), execute mirrors the same derivation onto the
returned copy; None stays None; the hand-seeded byte-identity dump pin
survives (absent -> skipped).
Gates: workspace 1037/0, clippy -D warnings clean.
refs #201
Task 1: RBootstrap and Generalization gain serde derives;
lineage.rs gains StageBootstrap (per_survivor | pooled_oos) and
CampaignGeneralization; StageRealization.bootstrap and
CampaignRunRecord.generalizations land serde-default sparse (a
pre-0108 campaign_runs.jsonl line still parses — pinned), threaded
None/empty through every literal site. No behaviour change.
Task 2: ExecFault::UnsupportedStage is gone; the preflight now admits
the v2 shape sweep [gate]* [wf]? [mc]? [generalize]? via a monotone
rank walk (precise per-pair PipelineShape prose) and statically
refuses: single-instrument generalize (GeneralizeNeedsInstruments),
non-R generalize metrics via check_r_metric (GeneralizeNonRMetric —
no fourth roster site), zero mc params (ZeroBootstrapParam). CLI
prose swapped accordingly. The loop's E2E phase added a tier-boundary
pin: process validate accepts [sweep, mc, walk_forward] while campaign
run refuses it data-free.
KNOWN RED (documented in the plan, flips in task 5):
crates/aura-cli/tests/research_docs.rs::campaign_run_v1_boundary_refuses_mc_process
still pins the retired 'not executable in v1' prose. Interim gates:
registry 57/0, campaign 33/0, cli --bin 111/0, workspace build clean.
refs #200
aura-engine: ListSpace — explicit point set implementing Space beside
GridSpace/RandomSpace (a gate's survivor subset has no cartesian
structure); GridSpace-identical arity/kind validation reusing the
existing SweepError variants (KindMismatch.value_index carries the
point ordinal for a list); empty point list is valid and sweeps to an
empty family; re-exported at the crate root.
aura-registry: CampaignRunRecord/CellRealization/StageRealization/
StageSelection — the thin campaign-level linking record over untouched
family records (#198 decision 4) — with append_campaign_run (assigns
run via next_campaign_run, a parallel counter beside next_run) and
load_campaign_runs (missing file => empty) on the campaign_runs.jsonl
sibling store; blueprint_path and find_blueprint_by_identity promoted
to pub (the campaign executor resolves the same refs the referential
tier validates).
Gates: engine 290/0, registry 54/0, clippy -D warnings clean.
refs #198
claude
Brummel