A source-blind consumer drove the shipped binary across recording and
re-recording, addressing a run after the fact, the description-edit authoring
loop, and a cold read of the documentation.
Both decisions this cycle made deliberately hold in every shape tested. A
parameter pinned on the command line to the value it already has gives the same
handle as not passing it, across one-at-a-time, both scalar kinds, and all at
once; the order of several `--override` flags never matters; a
description-only edit converges on the same directory. Eight invocations, three
directories, and the right three.
The field test also falsified two sentences this cycle wrote, both fixed here.
The contract clause claimed that two blueprints differing only in debug symbols
"compute bit-identically, so one address is the correct answer". They do compute
bit-identically — and are addressed separately, because `manifest.params`
records node-qualified names and the merged parameter vector is hashed. The
audit's substitution blanks those names inside the topology hash's slot; the
parameter vector re-admits them through a field it does not touch. Whether the
handle should follow the record or #171's name-blind projection is a genuine
fork between two established positions here, so the behaviour stands and the
clause now states what actually holds, naming #354 for the fork.
The authoring guide predicted two directories for a no-op `--override`, since
it listed "an `--override` value" among the splitting causes without mentioning
the merge. It now states the rule the cycle actually shipped — effective
parameterisation, not how it was supplied — and names the provenance exclusions
in the one place an author looks.
Routed: #354 (the rename fork), #355 (five friction items on the path from "a
run recorded something" to "I am looking at it" — handle discoverability,
`measure ic` refusal parity, asking for an identity without running, empty-state
answers, the README's command map), and the consumer-visible face of the
non-pruning class onto #352.
refs #311
Source-blind field test of the handle echo and the rewritten chart refusal
(range 9636b00..9f87e5a), driven from the public interface only: the built
binary's help and refusals, the authoring guide, the glossary, the ledger.
Replayable via c309_0_run_all.sh, which rebuilds its lab from scratch and
captures stdout/stderr per step.
The chain the cycle promised holds: take trace_name off a run's stdout, feed
it to chart and to measure ic, never once list the trace directory. Both legs
report the handle identically — including the measurement leg the audit
caught shipping without a pin — and a tap-free run emits no key at all,
consistently across stdout and the stored record. The family-id arm proved
its worth: the intuitive cut of a family id yields a REAL but wrong handle
(another run's data), and the arm refuses to make it.
Findings routed: the three bugs fixed in-cycle RED-first (51ac155 and its
successor); the friction and spec-gap items filed as a collective issue.
refs #309
Source-blind milestone fieldtest of the embedding promise (range
024e865..32f6be8): a path-dep World program driving refusals as values
and a live in-process tap, reproduce-as-data with both guard classes and
a tampered-metric divergence probe, 22 CLI exit-class probes derived
from the docs beforehand, and the stale-dylib handshake via the nodes-new
scaffold plus a rev-pinned external crate. Replayable via m4_0_run_all.sh.
Findings: 2 code bugs fixed in-cycle RED-first (99e9300 — the library
exposes-neither panic was the one real promise breach), 2 doc bugs fixed
in the same commit, 4 friction items bundled as #349, 2 spec gaps
ratified into C30 (rev+lockfile clause; the scaffold's absolute path also
in #349 — the committed lab-nodes fixture is relativized), 6 working.
refs #296
Source-blind per-cycle fieldtest over the sweep's four axes: the
declared-tap discovery loop closes in three commands without provoked
refusals; fixed{distance} is campaign-reachable with honest R
compression beside the vol regime; the C14 exit partition branches
reliably across sixteen probes; reference-semantics hashes agree
between record line and trace index across bare/no-op/substantive
overrides. Findings routed: both bugs fixed in-cycle (next commit),
frictions bundled as #345, the trace-overwrite spec gap consolidated
onto #311, the guide-rot instance evidenced on #344.
Milestone-scope consumer run on the release binary at 2f1bace, three
scenarios: cold start (help-only) reaches a first single run in 5
commands and a first selecting 4-cell family in 28 with 5 dead ends;
the five retired verbs refuse without re-routing; the full retired
research scope (grid, walk-forward, monte-carlo bootstrap, generalize)
runs document-first over two years of GER40+USDJPY in 2.4s with
bit-identical reproduce (21/21). A/B against the sugar-era cycle-300
transcript: comparable to first run, ~9x slower to first family,
strictly better artifact — the dominant cost is an introspect --unwired
bug on present-but-empty presentation (fixed in-cycle, follow-up
commit) plus the missing pre-run coverage view (evidence on the
tracker). 4 bugs / 6 friction / 3 spec-gap / 4 working, routed on the
tracker.
refs #319
Source-blind consumer run over the retired-sugar surface on the release
binary at 1476990: the blueprint smoke loop (op-script -> graph build ->
exec --tap/--override) runs and charts; twelve exec target-class probes
and the --override refusal family behave as documented on both legs; a
retired-verb consumer reaches a running 4-member real-data sweep family
through the document ramp; the campaign walk-forward zero-trade note
fires and stays silent correctly. The two bugs are exit-101 panics on
--override's blueprint leg (out-of-domain value; the retired wrapped
axis form) — fixed in-cycle RED-first in the follow-up commit. Friction
and spec-gap findings are bundled on the tracker; the trace-dir
overwrite evidence landed on the standing collision issue.
refs #319
Source-blind consumer run over the four cycle axes on the release
binary at 4ff85b9: the authoring loop lands the name in bytes, store
document, and traces/<name>/ (with the documented default collision
observable on the unnamed twin); the refusal battery is uniform and
op-indexed across the op route and the file intakes; a named blueprint
defaults its use-splice instance identifier as documented, two unnamed
splices collide, two named ones coexist; help/guide/glossary agree on
eleven ops and the render-name vs registry-label split. Finding B1
(sweep --list-axes ungated, mangled listing at exit 0) is fixed
RED-first in the preceding commit; SG1 (prefix variance in the refusal
prose) is ratified there in C24.
refs #331
Source-blind consumer fieldtest against the release binary at 4474814:
discovery-surface identity (--params vs --list-axes byte-identical,
raw + default= on bound), the document-first loop with two
bound-override axes (validate + run e2e, manifest raw), the translation
seams (all three verbs refuse wrapped identically exit 2; campaign
validate did-you-mean exit 1), and manifests raw in params AND defaults
on all routes with reproduce 4/4 bit-identical. One spec-gap: the guide
transcript abbreviated the strategy content-id (ratified in the
follow-up doc commit). Not exercisable source-blind: replay of a
pre-#328 wrapped-name on-disk family (noted in TRANSCRIPT).
refs #328
Milestone-close gate for milestone 37 (data-authorability boundary,
#310/#317/#271): source-blind consumer fieldtest against the release
binary at eb2b0a1, scenarios derived top-down from the milestone
promise rather than per-cycle axes.
- m37_1: NY Session authored via args + fold-selected taps, run e2e
over US500 (default and --tap plans; summaries correct, unlisted
tap inert).
- m37_2: London anchor registered by name, spliced via use into a
consumer with open SMA lengths, swept over GER40, reproduced 2/2
bit-identical (v2 propagates through the splice).
- m37_3a-e: five-way boundary-refusal battery; all exit 1 with
actionable prose, escalation path named for namespaced types.
Gate verdict: green — all three boundary mechanisms deliver the
promise; genuinely-new logic is turned away toward Rust. Findings
routed: fold-summary read-back absorbed into #309, cross-run tap-file
lifecycle evidence onto #311, bare-refusal escalation-pointer parity
onto #341 (refs #309, refs #311, refs #341).
Source-blind consumer run over the #271 surface: a NY first-bar-
momentum Session authored purely as data and run e2e over --real
US500; a LinComb{arity:3} blend binding the arity-unlocked weights
with the Session period swept (4 members, reproduced 4/4
bit-identically); eight strict-form refusal probes (all exit 1,
actionable prose); an args composite registered and use-spliced into
a consumer (v2 propagates, ids stable, C29 doc gate holds).
Findings routed: Count lexical form + asymmetry ratified into
glossary/guide (follow-up commit); introspect dead-end on arg-bearing
ports and the missing CostSum wiring path absorbed into #341.
refs #271
Source-blind per-cycle fieldtest of the #317 surface (register->label->
discover->use loop; ref forms + refusal UX; open patterns; sweep axes through
a spliced instance). Corpus under fieldtests/composites-use/ with TRANSCRIPT.md.
Working: the full authoring loop (W1), ref-form variety incl. ambiguous/
unknown-prefix and unknown-label refusals (W2), sweep --list-axes and a real
sweep through graph.<instance>.<node>.<param> plus splice-time bind (W3).
Findings, routed at close:
- B1 (reclassified: docs untruth, behaviour ratified): an open pattern whose
input role names an archive column RUNS standalone — the harness binds
roles by name, which is C26's design; the "running standalone refuses"
claim was the wrong sentence, fixed in the follow-up docs commit and
minuted on #317.
- S1: the run path's actual refusal trio (bias/tap gate, free knobs, role-
column) vs the docs' bootstrap-gate story -> same docs commit + #339 note.
- F1: gang-on-instance-member refusal does not name the unsupported-rule ->
evidence added to #339 item 1.
- S2: README still enumerated nine ops -> fixed in the follow-up commit.
- S3: the use-seam C29 gate is unreachable through today's CLI writes
(register gates both input forms) -> ratified as a backstop, one line in
C24.
- Orthogonal pre-existing observation (NOT #317): aura run --params did not
close free knobs in the fieldtester's hands -> #340.
refs #317
Source-blind against the release binary at 120d116, one example per
axis: hs_1_discovery (help -> --folds roster -> --tap spread=mean
without opening source or JSON; fold row at the last warm ts as
documented), hs_2_typo_key (three closed-set refusals — typo'd op key,
undeclared tap, unknown fold — each naming token + valid set, exit 1),
hs_3_three_taps (skipped-tap note fires exactly per skipped tap under
an explicit plan, silent under record-all and bind-all).
Findings routed: the op-key refusal cites the wrong line (points at
the next element; recovery still carried by the key name) -> #336;
declared-tap names have no positive discovery surface (the deferred
#333 view half, evidence now in hand) -> #337. All seven sweep
closures verified working from the outside.
refs #336, refs #337
Per-cycle fieldtest of the #310 --tap selector, source-blind against the
release binary (e482f0e). Spine holds: fold selection end-to-end
byte-exact incl. the C1 pin and kind-aware folds/refusals on a Bool tap;
parse refusals precise at exit 2. Findings routed to the tracker:
introspect --folds renders the aura-std FoldKind table instead of the
registry roster (capitalized ids, record missing); no public surface
enumerates a blueprint's declared taps; an explicit --tap plan drops
unlisted taps without the C14 'aura: note:' skipped-tap notice; fold
summary-row timestamp semantics undocumented and mixed. Exit-class
tension already tracked on #297 (audit).
refs #310
Binary-only milestone fieldtest for milestone 36 (self-description): a
fieldtester agent bootstrapped from the release binary (d26f0c8) alone —
no engine sources, no repo docs, no bootstrap card — and drove five
end-to-end scenarios: cold bootstrap -> author -> validate (green,
synthetic + real EURUSD, 3944 trades), execution semantics +
latch/edge-pulse idiom (green), vocabulary breadth (33 nodes / 7 folds /
17 metrics / 6+6 blocks, every entry with a meaning), document ramp from
a bare {} (green, 3 members), trace/measurement path (dead-ends,
tracked).
Verdict: the bootstrap card (#267, closed as superseded) is empirically
redundant for the bootstrap->author->validate spine. Two seams keep the
promise short of end-to-end, both tracked outside M1: the declared-tap
drop on the registration path (#327, the missing mechanism of #312/M6)
and the sugar<->document axis namespace (#328, prerequisite of the #319
retirement). Friction routed: #329#330#331; evidence comments on #312
and #324. Milestone 36 closed with this adjudication; the fieldtest spec
remains the usual git-ignored working file.
refs #267, refs #312, refs #319
All three mission drivers re-run against the rebuilt release binary
after 51096a3/a20b6f6/4897919: the synthetic poison row now shows
aura: warning: + exit 3 (zero raw panic lines in the survey), the
single-window case reads "the single walk-forward window recorded
zero trades", and the marker-separability log still greps into two
clean classes. Captures refreshed in place; the milestone fieldtest
roll-up is green. refs #278, refs #313
Consumer-seat run over all four C29 seams (load, register, op-script doc,
document description) from the public interface only. The three working
findings verify the #314 failure class removed end to end; the findings
route: guide sync + refusal-prose neutrality + doc_gate short-name parity
this run, discoverability surfaces to the tracker.
refs #316
Cycle-close fieldtest for #283 + #77 (tap subscribers), exercising the new
tap surface as a downstream consumer over the public interface only (design
ledger, glossary, cargo doc, example corpus — never the crate sources).
Five examples, one per axis, all built and run against worktree HEAD bd0c557:
- tap_1_record_spread.ops.json — record → runs/traces/graph/spread.json
- tap_4_measure_ic.ops.json — two-tap run → aura measure ic (clean IC)
- tap_consumer/src/tap_2_fold.rs — Named{mean}/Named{count} via run_signal_r
- tap_consumer/src/tap_3_live.rs — Live(closure); bit-identical to fold + record
- tap_consumer/examples/refusal_probe.rs — unknown-label refusal (exit 1, no write)
The consumer crate is a detached workspace (own empty [workspace], path-deps on
aura-core/-engine/-runner), exactly as a real World program; it is not a
workspace member and does not build under `cargo build --workspace`. Verified
here with a fresh `cargo build --bins --examples` (Finished, 0 errors).
No merge-blocking defect: record/fold/live/measure over the shared pair are
correct and deterministic. Follow-ups filed at cycle close:
- #309 (feature) — a plain `aura run` echoes no recorded trace handle (F1+SG1)
- #310 (idea) — Named fold has no data/CLI reach; ratify Rust-only vs selector (SG2)
- #311 (bug) — same-name re-run orphans still-chartable tap files (SG3, pre-existing)
- #297 (comment) — library refusal = process::exit, not Result (F2, already tracked)
Downstream-author fieldtest of the #300 surface, public interface only,
real GER40 archive. All four axes hold: the closed
generate -> show -> hand-extend -> re-register -> run loop (2-cell
extended campaign), the typed stop grammar's clap rejections, bare
plateau's content-id identity with plateau:mean, and the show read-back
(byte-identical, no framing) with a clean unknown-id refusal.
Findings routed: 1 bug (show outside a project blames the id instead of
the missing Aura.toml — RED-first fix follows), 2 friction items filed
to the tracker (opaque process-document select refusal; no prefix-id
hint on show), 4 working (carry-on). The fieldtest spec survives
git-ignored as the next planning cycle's reference input.
refs #300
Downstream-consumer corpus for the #295 library surface: a standalone
workspace (path-deps, no aura-cli in the link graph — cargo tree
verified) exercising the four cycle axes as a World program would,
public interface only: a DefaultMemberRunner member run over real
GER40, a hand-authored campaign document through
aura_campaign::execute, the IC + registry deflation dispatch as
library calls, and load_family/reproduce round-trips. Run transcripts
committed beside the sources.
Findings routed separately: the family-id mismatch between the
registry enumeration and the reproduce keying (list-then-reproduce
dead-ends) to debug; the reproduce-API friction (no report on the
simple path; DataSource/pip re-specification the manifest already
holds, wrong pip silently DIVERGED) to a follow-up. Four working
confirmations, chief among them the cycle's acceptance claim: all
four axes compiled first-try from the public docs alone.
refs #295
Milestone acceptance corpus for 'Measurement as a first-class citizen',
authored as a downstream consumer (public interface only): a gated
rate/mux/counter over the four new cells, a SessionFrankfurt anchor study,
and the origin first-candle-momentum study on real GER40 data — each a
hand-authored blueprint plus one recorded run, zero throwaway Rust. The
three capabilities compose end-to-end and record via declared taps.
Findings routed separately: two aura-run panics-where-refusal-expected
(bad tap wire; non-bias output) to debug; a tap-authoring friction and two
doc gaps to follow-ups.
Downstream-consumer fieldtest over the public surface at the close of the
maximize-throughput complex: a 4-instrument campaign authored, validated,
run and read back through the normal verbs; determinism probed across
--parallel-instruments 1/2/4/10 and a same-K repeat (byte-identical modulo
the run-counter suffix); duplicate-instrument refusal and faulted-cell
isolation (exit 3, healthy cells persist) confirmed under the parallel
loop. 5 working, 0 bugs, 1 friction (the zero-bound diagnostic leaks
clap's NonZeroUsize wording), 1 spec_gap (gate-emptied cell exit semantics
undocumented) — both resolved in the follow-up tidy.
refs #277
Per-cycle fieldtest over the #246 surface, consumer-perspective, public
interface only: the scaffold quickstart end to end (aura new -> run ->
--list-axes -> override sweep), --list-axes on a partially-open
blueprint, an override family reproduced bit-identically (3/3 members),
and the error surface (wrong axis name, kind-mismatched value — both
exit 2). Corpus under fieldtests/cycle-246-bound-override/.
Dispositions: the one bug (F6) is fixed in this commit — authoring-guide
§1 still asserted a bound param 'never shows up in --list-axes' and 'no
aura sweep --axis can reopen it', contradicting the shipped #246
semantics and the C12 amendment; the three-states paragraph now reads
open = must-bind, bound = overridable default. The friction finding (F5,
KindMismatch/MissingKnob printed as raw Debug structs beside polished
prose) is filed as #247.
refs #246
Consumer-POV fieldtest evidence for the two-tier project model: the
data-only quickstart, the role-2 attach loop, real-data usage inside a
data-only project, and the refusal surfaces all match the docs.
Inline fixes for two of the friction findings:
- docs/project-layout.md: the day-in-the-life run example now shows real
Unix-ms window bounds instead of bare years (--from/--to take ms).
- `aura new` --help one-liner no longer calls the project a crate.
Remaining findings tracked on the Gitea queue (empty first-manifest
provenance on a fresh scaffold; misleading no-data message for an
uncovered --real window; the tier hint missing on the graph-build path;
the raw cargo-metadata leak for a missing [nodes] dir).
refs #241
Milestone fieldtest for "Real-project readiness — role-complete
research loop (no freeze)": a from-scratch ger40-lab project (aura new)
with a brand-new native node (ger40_lab::Momentum) driven through the
full research loop on real GER40/FRA40 archive data — sweep,
walk-forward, mc, process+campaign with 2-regime risk and net-R cost
block, cross-instrument generalize, bit-identical reproduce, charts —
from the public interface only. Spine holds: 5 working, 1 bug (chart
cannot open a sweep --trace family by its printed handle), 4 friction,
1 spec gap. Findings triaged separately.
Consumer-perspective milestone fieldtest over the seven shipped items: ganged
authoring end to end (#61), consumer-blueprint render (#28), param-bearing
scaffold starter (#183), outside-project refusal (#218), default stops (#217),
gate-free plateau walkforward (#215), --trace honesty (#168) — all working.
Red half: aura reproduce DIVERGES 0/N on a dissolved-sweep (campaign-minted)
family and PANICS (main.rs:1123, non-empty OOS window) on a WalkForward
family — the C18/C1 reproduce promise breaks exactly on campaign-pipeline
families, invisible to per-cycle fieldtests. Both routed to debug this run.
Friction/gap findings filed as forward issues.
Fixture corpus: from-scratch ganged mid-channel op-script + built blueprint,
throwaway project, verbatim transcript. Report spec (git-ignored working
file): docs/specs/fieldtest-triage-harvest-milestone.md
refs #61
Milestone-close fieldtest for "Verb dissolution" (#210), authored from the
public interface only. Three end-to-end scenarios over the real GER40/USDJPY
2025 archive:
1. all four dissolved verbs on the r-sma real path (sweep via blueprint;
generalize/walkforward/mc via --strategy) — correct + deterministic;
2. reach-through — `aura generalize` auto-registers a campaign+process
document and `aura campaign run <id>` reproduces worst_case byte-for-byte;
3. a hand-authored 2-instrument × 2-window × 2-regime matrix (8 cells → 4
generalizations) no verb can express — regimes compared, never argmax'd.
Result: 0 bugs; the core promise holds empirically (behaviour preservation,
campaign-substrate reach-through, and the risk-regime axis all confirmed).
Residual findings are discoverability/ergonomics follow-ups, filed as #216
(risk-axis discoverability), #217 (verb knob asymmetry), #218 (no-project
store litter). The friction-and-bug spec stays a git-ignored working file.
Consumer scenario against the public surface only: selection-free
process doc round-trip, the dissolved real-data sweep (doc dedupe +
realization accumulation + C1 bit-identity), refusal edges, rank-later
over the persisted family. 0 bugs; the cycle's own axes all came back
working. 4 friction + 2 spec-gap findings on the verb entry ergonomics
and pre-existing guide drift — transcribed in
docs/specs/fieldtest-sweep-dissolution.md for triage.
refs #210
MILESTONE VERDICT: GREEN — the whole span (0106-0109) holds end to
end as one continuous consumer scenario: authored from a bare {} via
introspection alone, every ref resolved from the public surface, three
validation tiers, run by content id over two real instruments through
the full v2 pipeline, realization recorded (selection/survivors/
bootstrap/generalizations/trace_name), traces on disk and charted,
read back via campaign runs, byte-identical re-runs (C1), refusals
precise. 0 bugs, 0 frictions, 0 spec gaps; 2 doc-gaps.
F7 fixed in this commit (the docs claimed an archive-root path without
naming the key; a guessed [paths] archive silently no-ops): project-
layout and the glossary Aura.toml entry now name [paths] data, and the
aura new template advertises the commented-out key. F8 (op-script
grammar not CLI-discoverable) filed as #208, sibling of the #197 docs
half — one docwriter pass once desired.
The milestone-close gate evidence is this corpus; the formal milestone
close and the push remain user-reserved.
refs #189
Verdict: honest, recorded, reproducible inference on top of the 0107
loop — pooled_oos provably pools the wf OOS windows (n_trades 948+895
= 1843), per_survivor flags a razor-thin gate-passer at
prob_le_zero 0.525 (the inference layer earning its keep), the
two-instrument std::generalize records worst_case/sign_agreement over
last-wf-window nominees, byte-identical across three runs modulo the
run counter, and all eight boundary probes refuse loudly at the right
tier. 0 bugs.
Triage: F1-F5 carry-on; F6 (campaign validate blesses
executor-unrunnable shapes — the third tier invisible) and F11 (no
CLI read-back for recorded realizations) filed as issues; F7/F9/F10
doc-tightens follow this commit inline; F8 (generalize R-metric set
undiscoverable + misleading refusal prose for R-denominated-but-
unranked metrics) filed with the --metrics tag half. The fieldtest
spec rides in docs/specs/ as next-cycle planner input per convention.
refs #200
Downstream-consumer run (roles 5/6b, public interface only) over the
document-authoring loop: canonicalization byte-exact and stable across
reformatting (F8), intrinsic validation thorough and path-addressed (F9),
referential tier correct across the project boundary with progressive
resolution (F10). Findings: one bug (F2: introspect --block advertises an
identity_id process ref the validator refuses), four frictions (F1 envelope
not derivable from introspection incl. the block-id/field-name mismatch and
the un-blocked seed; F3 content:-prefixed id vs bare-hash refs breaking
referential resolution silently; F4 --unwired only reports omitted refs,
which validate then fails; F5 strategy refs unresolvable from the public
surface — no blueprint register/file-content-id/param-space verbs), one
doc-gap (F6 no worked examples, metric vocabulary unlisted), and the
spec-gap F7: a valid, registered campaign has no run path — the
intent-persistence re-test the executor decision waited on.
refs #189, #188
Milestone-close gate for 'Project environment — the project-as-crate
authoring loop' (#180), run as a downstream consumer from the public
interface only (README, project-layout, glossary, ledger, --help; no
crates/*/src, no crates/*/tests). Verdict: GREEN — 0 bugs.
Tasks (all ran, evidence in the corpus): (1) aura new -> cargo build ->
run twice byte-identical -> manifest project provenance
{namespace, dylib_sha256}; (2) a REAL authored node (Negate) in the
scaffolded lib.rs -> rebuild -> graph build resolves the project type ->
run flips total_pips sign, dylib_sha256 tracks the fresh build; (3)
renamed-twin id semantics: content-ids differ, identity-ids equal,
combinable output content-first, content-id == manifest topology_hash;
(4) misuse paths: charter violation named + refused exit 1,
missing --release build names path + fix, outside-project refused.
Findings (non-blocking, triaged separately): F1 friction (aura new git
init nests a gitlink under the engine tree), F2 doc-gap (no public
example of a bound-param build closure — the Identity template is
param-less), F3 doc-gap (glossary manifest entry lacks the project
provenance block), F4 friction (run leaks raw UnknownNodeType Debug
where graph build phrases house-style), F5 doc-gap (outside-project run
surfaces UnknownNodeType instead of a no-Aura.toml hint), F6 doc-gap
(scaffolded Aura.toml omits the documented data-archive-root key).
refs #180
Milestone-close validation of the World/C21 blueprint-data family surface, run
from the PUBLIC interface only (design ledger + op-script grammar; blueprints
authored as a consumer via `aura graph build`, no crates/ source read). Binary
built from 4710fcc.
The milestone's FUNCTIONAL PROMISE HOLDS (3 working findings): a consumer-authored
blueprint round-trips the whole verb family (graph build → list-axes → sweep →
rank → reproduce; walkforward → reproduce; mc → reproduce); all three FamilyKinds
(Sweep 4/4, MonteCarlo 5/5, WalkForward 3/3) reproduce bit-identically through the
one shared content-addressed store (the C18 core promise); list-axes→sweep→rank
composes and self-corrects.
Findings surfaced for follow-on (NOT fixed here — filed to the tracker): 2 bugs
(aura run on an open blueprint PANICS exit 101 instead of a clean exit-2 like mc;
closed→walkforward --axis prints the rejection twice), 2 friction (sweep-side
closed error terser than mc's exemplary message; mc reproduce lines omit the
seed), 2 spec-gaps (every open knob is mandatory on sweep but the docs frame
list-axes as a menu; the whole blueprint-data verb family is invisible from
`--help` / no README — empirically confirms #159). Detail per finding in the spec.
fieldtests/world-c21-milestone/ carries the op-scripts, authored blueprints,
captured outputs, and run_all.sh (regenerates the gitignored runs/ telemetry).
refs #170
#164 — tighten the C24 "canonical, versioned" blueprint contract on two corners
the milestone fieldtest surfaced. Both forks were derivable; resolved with
rationale (recorded on #164).
Fork 1 — canonical artifact carries no trailing newline. `blueprint_to_json`
returns the JSON value with no trailing newline (647 bytes for the SMA-cross
fixture); `aura graph build` had framed it with a `println!` (648). The library
serializer is the single canonical source — the form content-addressed topology
identity (#158) hashes — so the CLI is a transport that must not mutate the
bytes: `build_cmd` now `print!`s the canonical bytes verbatim, making the CLI and
library emit paths byte-identical. RED-first: the new E2E
`graph_build_emit_is_byte_canonical_no_trailing_newline` pinned the trailing
newline (FAIL), green after the one-line change. The two cycle-0088 `.out.json`
goldens are re-recorded at 647 bytes. The loader stays lenient (a trailing
newline on input still parses, Tier-1 robustness; the milestone fieldtest's
`mt_4` tolerance check is unaffected).
Fork 2 — the canonical JSON is the blueprint equality/identity surface; no
second in-memory PartialEq/Debug is added. `Composite`/`BlueprintNode` cannot
derive them: `PrimitiveBuilder` (aura-core node.rs) holds a
`build: Box<dyn Fn(&[Cell]) -> Box<dyn Node>>` closure, neither comparable nor
printable. Equality could only be defined over the serialized data, of which
`blueprint_to_json` is already the single source; a separate in-memory equality
notion would be a redundant second source of truth and a drift hazard against
the form #158 hashes. Recorded the decision (discharges acceptance box 2's "or
JSON-string is the intended equality surface" arm) in the C24 ledger; no code
added for it.
Verified: cargo test --workspace green (51 suites, incl. the new RED->green E2E);
cargo clippy --workspace --all-targets -D warnings clean.
closes#164
#163 — the construction op-script (`aura graph build`) had no durable public
schema doc, and the only worked-example corpus had drifted out of sync with the
current binary, so a consumer copying the visible example was misled.
Grammar reference (durable): a compact op-script grammar — the six verbs
(source/input/add/feed/connect/expose), their keys, the typed-Scalar bind form
({"I64":2}) and the capitalized ScalarKind, the by-identifier / dotted-port
model, and the add-`name` vs expose-`as` distinction — added to the C24 ledger
entry, with a correct worked-example pointer. Surfacing the same grammar in
`aura graph build --help` is left to ride with the CLI-discoverability work
(#159), keeping this change decoupled from the deferred World/project-as-crate
layer.
Corpus refresh: the cycle-0088 fixture corpus replayed against the current
binary failed on two accumulated drifts. (1) `add`'s naming key was renamed
`as` -> `name` (3c4b667; `expose` keeps `as` as a real alias) — the inputs now
use `name` for `add` ops only. (2) The dataflow-cycle bug the 0088 fieldtest
found was fixed (1652042, closes#161), so `c0088_3m_two_cycle` is now a
rejection fixture: the gap-era cyclic-blueprint golden is removed and
`c0088_3m.err` carries the eager cycle-rejection message. FINDINGS.md gets a
dated refresh stamp; its findings are preserved as written at fieldtest time.
Verified: the four golden-bearing success fixtures (_1, _2_completed,
_2_partial via --unwired) reproduce byte-identically; all 15 error fixtures
produce the intended by-identifier `op N (kind): cause` (or finalize) message.
Doc + fixtures only, no engine/CLI code touched.
closes#163
Closes the cycle-0090 (#156) work and delivers the "Topology-as-data:
blueprint serialization & loader (C24)" milestone.
Cycle-0090 audit (architect drift review, range 7ad7f58..HEAD): drift-clean
on substance. Tier-1 forward-tolerance is genuinely serde-default
(deny_unknown_fields = 0 in every production struct), pinned at envelope +
primitive across both the in-crate and public-seam tests; Tier-2 refusals
real and green; no struct/signature/runtime change; the invariant-5
acyclicity lockstep (construction gate <-> bootstrap Kahn sort) undisturbed;
invariant-9 resolver stays injected, no registry. The one medium item the
architect flagged — the C24 Remaining block carried no deferral signal for
#158/#159 — is fixed in this commit's ledger update.
Milestone fieldtest (fieldtests/milestone-topology-as-data/, the close-gate
evidence): GREEN, 0 behavioural bugs. A downstream consumer ran the full
author -> serialize -> load -> construct -> introspect -> reproduce story from
the public surface alone — round-trip bit-exact on both authoring surfaces
(Rust builder + op-script CLI), build-free introspection, Tier-1/Tier-2
forward-compat by name, fail-fast diagnostics by identifier. Verified the
headline claims myself (the two Rust binaries + the CLI build/diff + the
fail-fast exits), not just the agent's report.
Findings triaged to the forward queue (none blocking): op-script grammar
undocumented + a stale cycle-0088 example (#163); canonical trailing-newline
divergence + Composite Debug/PartialEq ergonomics (#164); CLI discoverability
of build/introspect + no `graph run` (commented on #159); the newline as a
content-addressing prerequisite (commented on #158).
#158 (content-address topology in the manifest) and #159 (retire the
hard-wired harnesses) were moved OUT of this milestone: both are premature
until the World generates runs from blueprint-data (commit still fully
identifies every hard-wired topology), so they belong to the
project-as-crate / World cycle. The C24 ledger Status records the delivery
and the deferral.
Ephemeral cycle-0090 spec + plan removed per the lifecycle convention.
The Gitea milestone container is closed on user ratification (this run
tees up the close, does not click it).
Exercise the construction op-script service (#157) purely as a downstream
consumer authoring topology as data: discover-then-author an SMA-crossover
bias, author-with-`--unwired`, and 15 error-ergonomics fixtures. Public
interface only (the C24 ledger + the CLI's own output); no implementation
source read.
Confirmed working: discover→author round-trip, the `--unwired` by-identifier
authoring aid, eager per-op diagnostics. Surfaced one bug + frictions, all
reproduced and filed forward:
- #161 (bug): `graph build` accepts a dataflow cycle — DAG invariant 5
unenforced; reproduced directly before filing.
- #162: construction error/bind ergonomics (raw-index finalize, Debug-struct
bind-mismatch, undiscoverable typed-Scalar bind).
- #28: no CLI consumer for the emitted #155 blueprint (round-trip dead-ends).
- #159: construction surface undiscoverable from the CLI usage/help.
FINDINGS.md carries the full report + the triage table. The two 1.4 MB
self-contained viewer HTML dumps from the round-trip check are dropped as
bloat (the byte-identical result is recorded in FINDINGS).
Milestone-close gate for "Cost-model graph (in R)" (#148). The fieldtester exercised
the milestone capability end-to-end from the public CLI only (no implementation
source), on the stage1-r harness, against ledger C10 + the glossary.
Roll-up: friction_found — 0 capability bugs; the milestone promise is empirically
delivered:
- bare gross-R run is clean (net == gross, no cost forced);
- per-trade cost: gross pinned, net drops strictly/monotonically, net_r_equity tap
persists;
- the capstone per-held-cycle carry BLEEDS the net-R curve continuously over a hold
(the gross-net gap grows 0.367 -> 0.734 -> 1.101 -> 1.468 across one hold, vs the
per-trade run's flat 0.367) — the holding-cost shape change, observable end-to-end;
- composition is exact-additive (the composed drag equals the sum of the single-cost
drags to the last ULP) and `aura chart` renders the real net_r_equity series.
Evidence under fieldtests/milestone-cost-model-graph/ (run_scenarios.sh +
captured_outputs.txt + traces/ + the charted bleed + FINDINGS.md).
Non-blocking findings filed forward (none gates the close):
- cost-flag CLI ergonomics — units hint, negative-rate diagnostic, non-R-harness
no-op + a C10 spec_gap -> #153;
- docs/project-layout.md describes the retired `aura backtest --broker` path and omits
the cost flags -> folded into #151.
refs #148
Milestone-close gate for "Inferential validation (defend against false
discovery at sweep scale)". Downstream-consumer exercise from the public
interface only (CLI --help, ledger, glossary, examples — never the source).
All four end-to-end scenarios PASS — the milestone delivers its promise:
trials-deflation (#144), plateau-over-peak (#145), and cross-instrument
generalization (#146) all reach the output and read sensibly, and the
composed "is this edge real or overfit?" verdict is honest (scenario 4: a
GER40 in-sample winner sqn 0.994 does NOT generalize — worst_case -0.149 on
EURUSD, sign_agreement 2/3 → instrument-specific / likely overfit).
Milestone status roll-up: bugs_found — NOT green, so the milestone is NOT yet
closeable (the green roll-up is the functional leg of the close gate). Findings:
- [bug] a family-emitting command piped into an early-closing reader
(`aura sweep|walkforward|mc|runs family … | head`) panics "Broken pipe
(os error 32)" on stdout (exit 101 under pipefail) instead of exiting
cleanly on EPIPE. Pre-existing (not a #146 regression — reproduced on
`aura sweep | head -1`); surfaced by the milestone commands. -> debug (RED-first).
- [friction] two generalize refusals (single --real symbol; multi-value
candidate flag) dump bare usage with no reason. -> tidy iteration.
- [friction] the --select plateau objective is absent from `aura --help`. -> tidy.
- [spec_gap] sweep vs generalize recompute the same member's sqn with a stable
1-ULP difference; the ledger's C1 / "recomputable" is silent on cross-command
metric agreement. -> ratify (ledger note).
- [spec_gap] the milestone vocabulary (deflated_score / overfit_probability /
worst_case / sign_agreement) is absent from docs/glossary.md. -> docwriter.
Artefacts: docs/specs/fieldtest-milestone-inferential-validation.md +
fieldtests/milestone-inferential-validation/ (README + 4 scenario scripts).
The existing fieldtests/ tree was untouched.
refs #146
Milestone-close fieldtest for "Runway — real-data ergonomics & honesty
hardening". Four curated end-to-end scenarios derived top-down from the
milestone promise, run as a downstream consumer against real GER40/EURUSD M1
bars (public interface only; no crates/*/src read):
1. per-instrument pip honesty (#22) — GER40 index pip 1.0 AND EURUSD forex
pip 0.0001 both honest end-to-end; AAPL.US/FOOBAR un-vetted refused
exit 2, no stdout leak
2. real GER40 M1 open via open_ohlc from aura-ingest ALONE (#80/#81/#92) —
no data-server dep, 53173 real bars, C4 O/H/L/C order, ts-window inverse
3. multi-tap ts-join (#93) — two taps at different cadences, 27744 rows,
0 by-ts mismatches; warm-up None rows where a positional zip would corrupt
4. run-registry coherence (#73/#82) — list/rank retired exit 2, families/
family live, two-store Registry read-back + rustdoc
Findings: 4 working, 1 friction, 1 spec_gap.
- friction: Cell exposes only from_i64/from_f64/... while the side-by-side
Scalar is a tagged enum with public I64(..)/F64(..) variants. The natural
param-authoring entry point is Scalar (Cell is the hot-path buffer carrier),
so the wrong reach is a self-correcting compile snag — ergonomic only.
Filed as a follow-up idea.
- spec_gap: unknown family id exits 0 empty — ratified into the C18 note in
the preceding commit (treat-as-empty contract).
Gate verdict: GREEN. All seven axes mapped to a covering scenario and exercised
on real bars; both finding classifications and the one material coverage gap
were adversarially re-verified. That gap — #22's FX-vetted pip face, untested by
the original scenario set — was probed (EURUSD -> sim-optimal(pip_size=0.0001),
FX-scaled, not index-mis-scaled) and folded into scenario 1, so the pip promise
is demonstrated on both positive faces. Every scenario re-run independently on
HEAD before commit.
The residency probe was an inherent method on the concrete M1FieldSource, so a Box<dyn Source> consumer could not read it without downcasting (the adjacent friction noted in #95). Hoist it onto the Source trait as resident_records(&self) -> Option<usize> with a default of None, mirroring bounds()'s Option-for-unknown idiom: None = "does not report" (the eager VecSource, which has no streaming ring), distinct from Some(0) = "reports, holds none" (e.g. exhausted).
M1FieldSource overrides it to return Some(chunk len). Callers updated (streaming_seam, the two deep-dive fieldtest bins). Tests: a hermetic VecSource default-is-None check, and a &dyn Source probe in the gated residency test proving M1FieldSource residency is readable without a downcast.
refs #95
Public-API research deep-dive driving the just-shipped GER40 real-data
session-breakout demo (5b5f034) through the escalating sequence a serious
researcher attempts next: scale to multi-year, walk-forward, structural-param
sweep, multi-instrument compare. Standalone consumer crate (path-deps only;
public interface = ledger + glossary + example corpus + cargo doc; no
crates/*/src read). All runs against the real /mnt/tickdata/Pepperstone archive.
Findings: 3 bugs, 1 friction, 3 spec_gap, 1 working. Triaged to the tracker:
- spec_gap: the flagship breakout ships as a raw FlatGraph the World API cannot
consume (sweep / walk_forward / compare all forced a hand re-author to a
Composite, which reproduced the FlatGraph numbers EXACTLY) — refs #94.
- bug: process residency is O(window) (6->173 MiB across 1mo->10y) despite the
documented O(one-chunk) claim; the aura ring stays <=1024 but data-server
retention scales, and streaming_seam.rs only measures the ring — refs #95.
- bug: the bar-period is one knob split across Resample (tunable param) and
Session (baked) and desyncs silently to 0.0 pips; Delay.lag leaks into the
default param_space (the C34 deform-vs-tune discriminator) — refs #96.
- spec_gap: the walk_forward empty-space idiom for param-free strategies is
undefined — refs #97.
- spec_gap (no per-instrument pip registry) and friction (one-directional
ns->ms conversion) re-confirm the known #22 and #80.
- working: C1 determinism over 7y / 148k bars, the lazy streaming source, and
the sweep/WFO cores held under real multi-year load; the Composite re-author
was behaviour-preserving (C23).
Scratch consumer crate + run transcripts committed under fieldtests/ per the
fieldtest convention (matching cycle-0049). The crate path-deps the engine and
reads the real archive; its bins are the repros cited in the issues above.
Milestone-close gate for "The World, part II — orchestration families".
Four end-to-end scenarios derived top-down from the milestone promise,
each built from HEAD and run as a downstream consumer through the public
API + CLI only (no crates/*/src read):
- mw_3 (headline): real EURUSD-M1 walk-forward composing all four axes
(grid sweep + optimize in-sample, OOS stitch) over the M1FieldSource
seam + lineage — 5 windows, 145 200-pt stitched curve, C2 held.
- mw_2: monte_carlo over 128 seeds, reproducible McFamily (C1) + aggregate.
- mw_1: random sweep -> optimize -> 64-member lineage round-trip.
- mw_4: family-lineage CLI transcript.
Findings: 4 working (promise empirically delivered), 1 bug, 3 friction.
- bug F5: `aura runs list`/`rank` cannot read a pre-0047 runs.jsonl — the
cycle-0047 typed-Scalar params migration (86746e3) left the persisted
flat store unmigrated; load() fails at the first bare float (+ exits 0).
- friction F6: no epoch_ns_to_unix_ms inverse for real-data walk-forward.
- friction F7: DataServer/DEFAULT_DATA_PATH not re-exported by aura-ingest.
- friction F8: family store is directory-keyed, undocumented on Registry::open.
Spec: docs/specs/fieldtest-milestone-the-world-orchestration-families.md
Public-API field test of the random param-sweep surface, from a standalone
downstream-consumer crate (path-deps only; the public interface = ledger +
glossary + spec 0049 + cargo doc rustdoc; no crates/*/src read). Four bins,
each built from HEAD and run: continuous tuning (200-point random tune ranked
by total_pips), the typed validation gate (all five reachable SweepError
variants pre-run), reproducibility + seed-sensitivity + the full i64::MIN..=MAX
sampler edge, and Space-trait interchangeability (one tune_and_rank<S: Space>
over both GridSpace and RandomSpace).
Findings: 0 bugs, 4 working, 2 spec_gap, 1 friction. The four working findings
confirm the cycle's acceptance criterion empirically — the headline tune reads
as the code a researcher would write, the gate is precise and fires before any
run, the C1 reproducibility promise is checkable in one line, and the Space
trait delivers one-consumer/both-enumerations.
Triage of the actionable findings:
- friction (no named-axis builder for RandomSpace — positional Vec<ParamRange>
must align with param_space() by hand, and a same-kind transposition passes
validation silently): filed as a feature for a future cycle, refs #79
(a RandomBinder sibling to the grid's SweepBinder).
- spec_gap (SweepError rustdoc summary named only GridSpace): fixed inline in a
follow-up doc commit.
- spec_gap (NonNumericRange / Bool-slot ranges unreachable with the shipped
aura-std node roster — no node declares a Bool/Timestamp knob): RATIFIED as
intentional. The variant is a forward-looking structural guard for the C16
"author your own node" path (a Bool/Timestamp param-slot a custom node may
declare); its current untriggerability with the standard roster is expected,
not drift.
refs #79
Per-cycle fieldtest of node-instance naming (#56), driven as a downstream
consumer from the public interface only (ledger + spec + rustdoc + CLI; no
crates/*/src read), three fixtures built and run from HEAD.
Verdict: the core 0031 promise holds first-try. A consumer authors
Sma::builder().named("fast"), inspects param_space() (sma_cross.fast.length),
binds by name and runs; the default-name case (sma.length / exposure.scale,
verbatim lowercase) and paramless-interchangeable-stays-legal both hold.
0 bugs, 3 working, 2 friction, 1 spec_gap. The one real gap (verified by the
orchestrator against the fixture output): the spec's headline forcing function
IndistinguishableFanIn does NOT reach an author on the canonical
.with(...).bootstrap() flow. An un-named 2-SMA cross emits a literal DUPLICATE
knob (sma_cross.sma.length x2); the binder resolves names before the compile
fan-in check, so the author hits UnknownKnob / AmbiguousKnob("sma_cross.sma.length")
— which point at the knob, not at the cure "name your nodes". IndistinguishableFanIn
only surfaces via the positional compile_with_params path. Rejection still
happens (no invalid blueprint runs), so it is an ergonomic/signposting gap, not
a correctness bug. Routed to the backlog (relates to #58); 0031 stays
audit-closed. Minor: FlatGraph/Harness lack Debug, so a bootstrap Result can't
be {:?}-printed.
Milestone-scope fieldtest (closing functional gate) for "The World —
parameter-space & sweep" (issues #30-#36). Three curated end-to-end scenarios
derived top-down from the milestone promise, driven from the public interface
only (ledger + specs + rustdoc; no crates/*/src read), built and run from HEAD
via a downstream consumer crate.
Delivery verdict: GREEN. The milestone delivers its promise.
- mw_1 single run bound by name: .with("sma_cross.fast", 2).bootstrap() binds
and runs (12 equity rows); diagnostics precise and knob-named (UnknownKnob /
AmbiguousKnob / KindMismatch).
- mw_2 named-axis sweep + compare: .axis(..).sweep(run) yields a 4-member,
ordered (odometer), disjoint, non-constant SweepFamily; ranking from public
fields. CLI (aura sweep / aura runs rank) delivers the same with no Rust.
- mw_3 structural-constant negative space (deliberate): SMA2-entry's
definitional `2` has no honest expression today -> grounds #55. Out of
milestone scope; not a blocker.
Findings: 0 bugs, 2 working, 3 friction, 1 spec_gap. Verified independently
(re-ran all three from HEAD; confirmed a plain SMA-cross emits sma_cross.length
twice and the shipped CLI sample carries the fast/slow ParamAlias overlays).
Friction (routed to the forward queue, none blocks the gate):
- the promised sma_cross.fast/.slow names are not free — they require
author-added ParamAlias overlays, doubly mandatory because the canonical
2-SMA cross also will not bootstrap without them (IndistinguishableFanIn);
- the sweep closure drops back to positional bind + manual name re-zip.
spec_gap -> #55: the wished-for structural-constant consumer code is recorded
in mw_3 as #55's acceptance evidence.
This commit is the milestone fieldtest itself; it closes no issue. Closing the
tracker milestone stays a deliberate manual act on this green gate.
Brings the other two composite-boundary edge-kinds to the named-projection
shape #40 gave outputs. input_roles changes from a bare Vec<Vec<Target>>
to Vec<Role { name, targets }> (rendered [in:<name>]); a composite gains
params: Vec<ParamAlias { name, node, slot }> that relabels an interior
leaf param slot's surface name in param_space() (rendered [param:<name>]).
Param aliasing is a PURE NAMING OVERLAY, not curation (the load-bearing
decision, per spec 0019): every interior param slot stays in param_space()
and sweepable; the alias only relabels in place, never reorders or hides.
Proven empirically — the MACD run is byte-identical (total_pips
0.1637945563898923, 3 sign flips), only the param labels improved:
param_space() now surfaces [macd.fast, macd.slow, macd.signal] instead of
three indistinguishable macd.length, and the manifest reads ema_fast/
ema_slow/ema_signal.
C23 honoured: role/param/output names are non-load-bearing debug symbols
dropped at lowering; identity is positional (role index, param slot,
output field). compiled_view_golden is byte-identical (verified: the
golden region is untouched in the diff). An out-of-range alias (missing/
non-leaf node or slot past the leaf's param count) is rejected at
compile_with_params as BadInteriorIndex, mirroring the output range-check
(no new variant). Orthogonal to #36 — purely additive at the composite
level.
Aliasing is demonstrated on the CLI MACD site only (the spec's worked
example + a new E2E test macd_param_space_surfaces_the_three_named_aliases);
sma_cross, the engine test fixtures, and the construction-layer fieldtests
get the forced role-name + empty params, so the param_space C23 anchor
goldens (param_space_mirrors_compiled_flat_node_param_order + siblings)
stay byte-identical.
Verification (orchestrator-run, not trusted from the agent report):
cargo build/test/clippy --workspace -D warnings all green (engine 66,
cli 12); the separate-workspace construction-layer fieldtest crate builds
(guards the #42 latent-drift recurrence); compiled_view_golden + MACD
determinism unchanged.
Two faithful repairs to the plan's literal test/code bodies, no semantic
change: the out-of-range test uses .err()/Some(BadInteriorIndex) (the Ok
arm Vec<Box<dyn Node>> is not Debug, so .unwrap_err() would not compile),
and the inline_composite destructure binds params as `param_aliases` to
avoid shadowing the injected `params: &[Scalar]` arg.
closes#41
The four standalone construction-layer fieldtest bins (mc_1..mc_4) no
longer compiled: they are their own workspace root, so the engine's
`cargo build --workspace` never touched them and two waves of API drift
went latent.
Two breakage classes, both confined to the fixture files (no engine
change):
- Stale cycle-0016 authoring. `BlueprintNode::from(<Node>::new(..))`
relied on the `From<NodeType>` impls retired in the value-empty
migration; only `From<LeafFactory>` survives. Ported to
`<Node>::factory().into()` — nodes are value-empty blueprint items,
params bound at compile. Knock-on in mc_4 `describe_node`: a blueprint
leaf is now a `&LeafFactory` with no pre-build schema, so leaf
introspection reports the render label + declared param count (inputs
/ outputs exist only on the compiled flat node, post-build).
- Cycle-0018 (#40) OutPort -> OutField slice drift. `Composite::output()`
now returns `&[OutField]`; the mc_4 read sites were still treating it
as a single port. Ported to index the record; the walk assertion now
also pins the arity (one re-exported field), which is the #40 shape.
Faithful ports, not assertion-gutting: each fixture still exercises its
axis (composite build+run / miswire render / nested composite / graph
introspection), verified by running all four bins to their `OK:` line.
One genuine semantic drift flagged in mc_2: the blueprint-view label is
now the bare value-empty type ("SMA"), so two same-type leaves share a
label and are disambiguated by the slot/edge table; the valued
SMA(2)/SMA(4) label lives on the compiled view. The fixture asserts what
the blueprint view actually exposes and keeps the load-bearing edge-table
observability check untouched.
closes#42
Replace a composite's single output port (OutPort { node, field }) with
an ordered, named record (Vec<OutField { node, field, name }>). A
composite can now re-export K interior fields as one output; a consumer
selects one via the existing Edge::from_field — the same field-wise
selector that already works for multi-output leaf producers (OHLCV).
Why: multi-line indicators (MACD, Bollinger, Stochastic, Ichimoku) could
not be authored as a composition and re-exported as a unit — only one
line escaped the boundary. The MACD PoC was forced to expose only the
histogram. This lifts the three `field == 0` / `from_field == 0` caps at
the composite boundary (inline_composite nested arm, rewrite_edge
composite arm), range-checking the index instead.
Boundary completion, not an invariant change:
- C8/C7/C4 untouched — one port, one row per eval, K base columns (a
3-line MACD is identical in kind to OHLCV).
- C23 honoured — re-export names live at the blueprint boundary only and
are dropped in the compilat (raw index wiring). compiled_view_golden
stayed byte-identical (the regression guard): no name leaked into the
flat graph.
The MACD PoC now re-exports macd / signal / histogram as a record; the
strategy still trades the histogram, selected via from_field: 2. The
render shows K [out:<name>] markers per multi-output composite (input
roles stay [in:<index>] — naming them is the dependent #41).
Output naming ships with the capability (OutField is born name-ready) so
#41 — composite param aliasing + input-role naming — does not reopen the
type.
Verification: cargo build/test/clippy --workspace -- -D warnings all
green; aura-engine 62 tests (+3 capability, +1 through-run E2E), aura-cli
unchanged-count green; compiled_view_golden byte-identical.
Known debt (out of scope, pre-existing): the non-workspace construction-
layer fieldtests (fieldtests/milestone-construction-layer/mc_1..mc_4)
carry their OutField sweep but still do not compile standalone — they use
the Sma::new / BlueprintNode::from(Sma) API retired in the cycle-0016
value-empty migration, never propagated to these fixtures. Tracked as a
follow-up.
closes#40
claude
Brummel