devrites 6.0.4 → 6.1.0

package/engine/internal/overhaul/score/run_test.go

@@ -1,636 +0,0 @@
-package score
-
-import (
-	"bytes"
-	"encoding/json"
-	"fmt"
-	"maps"
-	"math/big"
-	"os"
-	"path/filepath"
-	"strings"
-	"testing"
-
-	"github.com/devrites/devrites/internal/overhaul/ovio"
-)
-
-var doms = []string{"correctness", "security", "reliability", "performance", "tests", "architecture", "ux", "operations"}
-
-var fp = strings.Repeat("a", 64)
-
-func ctl(i int, d string, w any, lanes ...string) map[string]any {
-	if len(lanes) == 0 {
-		lanes = []string{"backend"}
-	}
-	return map[string]any{"id": fmt.Sprintf("C-%03d", i), "domain": d, "lanes": lanes, "weight": w, "hard_gate": false, "evaluation": "executed"}
-}
-
-func rubricOf(controls []map[string]any) map[string]any {
-	return map[string]any{"schema": "overhaul.rubric/1", "rev": 1, "lanes": []string{"frontend", "backend"},
-		"controls": controls, "domains": map[string]any{"correctness": 20, "security": 20, "reliability": 15,
-			"performance": 15, "tests": 10, "architecture": 10, "ux": 5, "operations": 5}}
-}
-
-func resOf(status map[string]string) map[string]any {
-	r := map[string]any{}
-	for k, v := range status {
-		r[k] = map[string]any{"status": v, "evidence": []string{"EV-1"}}
-	}
-	return map[string]any{"schema": "overhaul.results/1", "subject": "candidate", "fingerprint": fp, "results": r}
-}
-
-func passing() map[string]any {
-	g := map[string]any{}
-	for _, id := range auditGates {
-		g[id] = map[string]any{"status": "PASS", "evidence": []string{"EV-g"}}
-	}
-	return map[string]any{"mode": "audit", "gates": g}
-}
-
-func allPass(cs []map[string]any) map[string]string {
-	st := map[string]string{}
-	for _, c := range cs {
-		st[c["id"].(string)] = "PASS"
-	}
-	return st
-}
-
-// norm round-trips v through JSON with UseNumber, exactly as file input arrives.
-func norm(t *testing.T, v any) map[string]any {
-	t.Helper()
-	b, err := json.Marshal(v)
-	if err != nil {
-		t.Fatal(err)
-	}
-	dec := json.NewDecoder(bytes.NewReader(b))
-	dec.UseNumber()
-	var m map[string]any
-	if err := dec.Decode(&m); err != nil {
-		t.Fatal(err)
-	}
-	return m
-}
-
-func tryScore(t *testing.T, r, s, g any) (map[string]any, error) {
-	t.Helper()
-	out, err := score(norm(t, r), norm(t, s), norm(t, g))
-	if err != nil {
-		return nil, err
-	}
-	return norm(t, out), nil
-}
-
-func mustScore(t *testing.T, r, s, g any) map[string]any {
-	t.Helper()
-	out, err := tryScore(t, r, s, g)
-	if err != nil {
-		t.Fatalf("score: %v", err)
-	}
-	return out
-}
-
-func get(v any, path string) string { return fmt.Sprint(ovio.Get(v, path)) }
-
-func exact(t *testing.T, v any, path string) *big.Rat {
-	t.Helper()
-	r, ok := new(big.Rat).SetString(get(v, path))
-	if !ok {
-		t.Fatalf("%s is not exact: %v", path, ovio.Get(v, path))
-	}
-	return r
-}
-
-func base() []map[string]any {
-	var b []map[string]any
-	for i, d := range doms {
-		b = append(b, ctl(i, d, 5, "frontend", "backend"))
-	}
-	return b
-}
-
-func workedExample() ([]map[string]any, map[string]string) {
-	ex := append(base(), ctl(10, "correctness", 5), ctl(11, "correctness", 5), ctl(12, "correctness", 3), ctl(13, "correctness", 1))
-	st := allPass(ex)
-	st["C-013"] = "FAIL"
-	return ex, st
-}
-
-func laneMasking() []map[string]any {
-	l := base()
-	for k := range 10 {
-		l = append(l, ctl(30+k, "correctness", 5, "backend"))
-	}
-	return append(l, ctl(40, "correctness", 5, "frontend"))
-}
-
-// 1. Worked example: correctness 5,5,5,3 PASS + 1 FAIL -> D=180/19, Q=18800/1900.
-func TestWorkedExample(t *testing.T) {
-	ex, st := workedExample()
-	out := mustScore(t, rubricOf(ex), resOf(st), passing())
-	if got := get(out, "global.domains.correctness.exact"); got != "180/19" {
-		t.Fatalf("correctness = %s", got)
-	}
-	if got := get(out, "global.Q.exact"); got != "188/19" {
-		t.Fatalf("Q = %s", got)
-	}
-	if got := get(out, "global.Q.decimal"); got != "9.894736842105" {
-		t.Fatalf("decimal = %s", got)
-	}
-	if v := get(out, "readiness_verdict"); v != "MEETS_TARGETS" {
-		t.Fatalf("verdict = %s", v)
-	}
-}
-
-// 2. Same numbers, failing control is a hard gate -> NOT_READY despite Q 9.89.
-func TestHardGateBlocks(t *testing.T) {
-	ex, st := workedExample()
-	ex[len(ex)-1]["hard_gate"] = true
-	out := mustScore(t, rubricOf(ex), resOf(st), passing())
-	if get(out, "readiness_verdict") != "NOT_READY" || get(out, "gates.G-MANDATORY") != "FAIL" {
-		t.Fatalf("verdict %s, G-MANDATORY %s", get(out, "readiness_verdict"), get(out, "gates.G-MANDATORY"))
-	}
-	if got := get(out, "global.Q.exact"); got != "188/19" {
-		t.Fatalf("Q = %s", got)
-	}
-}
-
-// 3. Q exactly 9.696 fails 9.7 at full precision.
-func TestFullPrecisionThreshold(t *testing.T) {
-	var cs []map[string]any
-	st := map[string]string{}
-	i := 100
-	for _, d := range doms {
-		n := 1
-		if d == "correctness" || d == "security" {
-			n = 250
-		}
-		for k := range n {
-			cs = append(cs, ctl(i, d, 1, "frontend", "backend"))
-			st[fmt.Sprintf("C-%03d", i)] = "PASS"
-			if n == 250 && k < 19 {
-				st[fmt.Sprintf("C-%03d", i)] = "FAIL"
-			}
-			i++
-		}
-	}
-	out := mustScore(t, rubricOf(cs), resOf(st), passing())
-	if got := get(out, "global.Q.exact"); got != "1212/125" {
-		t.Fatalf("Q = %s", got)
-	}
-	row := ovio.List(out["thresholds"])[0]
-	if get(row, "scope") != "global" || get(row, "result") != "FAIL" || get(out, "readiness_verdict") != "NOT_READY" {
-		t.Fatalf("row %v verdict %s", row, get(out, "readiness_verdict"))
-	}
-}
-
-// 4. Zero denominator: an applicable domain whose only control is N/A -> NOT_ASSESSABLE, never 10.
-func TestZeroDenominatorNotAssessable(t *testing.T) {
-	b := base()
-	z := make([]map[string]any, len(b))
-	for i, c := range b {
-		z[i] = maps.Clone(c)
-	}
-	z[6]["na"] = map[string]any{"reason": "no UI", "evidence": []string{"EV-2"}, "approved": true}
-	out := mustScore(t, rubricOf(z), resOf(allPass(b)), passing())
-	if ovio.Get(out, "global.Q") != nil || get(out, "readiness_verdict") != "NOT_ASSESSABLE" {
-		t.Fatalf("Q %v verdict %s", ovio.Get(out, "global.Q"), get(out, "readiness_verdict"))
-	}
-}
-
-// 5. Global 59/6 and backend 10 cannot mask frontend 9.0; shared controls count once globally.
-func TestLaneMasking(t *testing.T) {
-	l := laneMasking()
-	st := allPass(l)
-	st["C-040"] = "FAIL"
-	out := mustScore(t, rubricOf(l), resOf(st), passing())
-	if row := ovio.List(out["thresholds"])[0]; get(row, "result") != "PASS" || get(out, "global.Q.exact") != "59/6" {
-		t.Fatalf("global row %v Q %s", row, get(out, "global.Q.exact"))
-	}
-	if get(out, "lanes.frontend.Q.exact") != "9/1" || get(out, "lanes.backend.Q.exact") != "10/1" {
-		t.Fatalf("lanes %v", out["lanes"])
-	}
-	laneRow := "missing"
-	for _, row := range ovio.List(out["thresholds"]) {
-		if get(row, "scope") == "lane frontend" {
-			laneRow = get(row, "result")
-		}
-	}
-	if laneRow != "FAIL" {
-		t.Fatalf("lane frontend threshold row = %s", laneRow)
-	}
-	if get(out, "readiness_verdict") != "NOT_READY" {
-		t.Fatalf("verdict %s", get(out, "readiness_verdict"))
-	}
-	if get(out, "counts.controls") != "19" || get(out, "counts.lane_appearances") != "27" {
-		t.Fatalf("counts %v", out["counts"])
-	}
-}
-
-// 6. PASS without evidence counts as UNKNOWN; UNKNOWN earns zero, not N/A.
-func TestPassWithoutEvidenceIsUnknown(t *testing.T) {
-	b := base()
-	r := resOf(allPass(b))
-	r["results"].(map[string]any)["C-000"].(map[string]any)["evidence"] = []string{}
-	out := mustScore(t, rubricOf(b), r, passing())
-	if get(out, "status_by_control.C-000") != "UNKNOWN" || get(out, "global.domains.correctness.exact") != "0/1" {
-		t.Fatalf("status %s correctness %s", get(out, "status_by_control.C-000"), get(out, "global.domains.correctness.exact"))
-	}
-	if len(ovio.List(out["downgrades"])) != 1 {
-		t.Fatalf("downgrades %v", out["downgrades"])
-	}
-}
-
-// 7. Fixed-catalog monotonicity: turning any PASS into FAIL never raises any score.
-func TestMonotonicity(t *testing.T) {
-	l := laneMasking()
-	full := allPass(l)
-	top := mustScore(t, rubricOf(l), resOf(full), passing())
-	for id := range full {
-		worse := maps.Clone(full)
-		worse[id] = "FAIL"
-		low := mustScore(t, rubricOf(l), resOf(worse), passing())
-		for _, p := range []string{"global.Q.exact", "lanes.frontend.Q.exact", "lanes.backend.Q.exact"} {
-			if exact(t, low, p).Cmp(exact(t, top, p)) > 0 {
-				t.Fatalf("failing %s raised %s", id, p)
-			}
-		}
-		if exact(t, low, "global.Q.exact").Cmp(exact(t, top, "global.Q.exact")) >= 0 {
-			t.Fatalf("failing %s did not lower global Q", id)
-		}
-	}
-}
-
-// Adding a FAIL or UNKNOWN control strictly lowers its own domain and never raises any domain.
-func TestAddedNonPassControlLowersItsDomain(t *testing.T) {
-	b := base()
-	top := mustScore(t, rubricOf(b), resOf(allPass(b)), passing())
-	for _, d := range doms {
-		for _, status := range []string{"FAIL", "UNKNOWN"} {
-			t.Run(d+"/"+status, func(t *testing.T) {
-				cs := append(base(), ctl(50, d, 1, "frontend", "backend"))
-				st := allPass(cs)
-				st["C-050"] = status
-				out := mustScore(t, rubricOf(cs), resOf(st), passing())
-				for _, o := range doms {
-					p := "global.domains." + o + ".exact"
-					cmp := exact(t, out, p).Cmp(exact(t, top, p))
-					if o == d && cmp >= 0 {
-						t.Fatalf("%s control did not lower %s: %s", status, d, get(out, p))
-					}
-					if cmp > 0 {
-						t.Fatalf("%s control in %s raised %s", status, d, o)
-					}
-				}
-			})
-		}
-	}
-}
-
-// A control shared by two lanes adds its weight to the global domain denominator once.
-func TestSharedControlWeightCountsOnceGlobally(t *testing.T) {
-	cs := []map[string]any{ctl(0, "correctness", 5, "frontend", "backend"), ctl(1, "correctness", 5, "backend")}
-	for _, d := range doms[1:] {
-		cs = append(cs, ctl(2+len(cs), d, 5, "frontend", "backend"))
-	}
-	st := allPass(cs)
-	st["C-001"] = "FAIL"
-	out := mustScore(t, rubricOf(cs), resOf(st), passing())
-	if got := get(out, "global.domains.correctness.exact"); got != "5/1" {
-		t.Fatalf("correctness = %s, want 5/1", got)
-	}
-}
-
-// 8. Duplicate control ids, weights off the 1/3/5 scale and unknown evaluations are invalid input.
-func TestInvalidCatalog(t *testing.T) {
-	zero := base()
-	zero[0] = maps.Clone(zero[0])
-	zero[0]["weight"] = 0
-	two := base()
-	two[0] = maps.Clone(two[0])
-	two[0]["weight"] = 2
-	noEval := base()
-	noEval[0] = maps.Clone(noEval[0])
-	delete(noEval[0], "evaluation")
-	for _, tc := range []struct {
-		cs   []map[string]any
-		want string
-	}{{append(base(), base()[0]), "duplicate control id"}, {zero, "must be > 0"}, {two, "weight must be 1, 3 or 5"},
-		{noEval, "evaluation must be"}} {
-		if _, err := tryScore(t, rubricOf(tc.cs), resOf(nil), passing()); err == nil || !strings.Contains(err.Error(), tc.want) {
-			t.Fatalf("want %q, got %v", tc.want, err)
-		}
-	}
-}
-
-// 9. Approved sequential degradation is the only open gate -> qualified outcome.
-func TestApprovedDegradation(t *testing.T) {
-	b := base()
-	g := passing()
-	g["gates"].(map[string]any)["G-CONCURRENCY"] = map[string]any{"status": "WAIVED_OPERATIONAL", "approval": "AP-2"}
-	out := mustScore(t, rubricOf(b), resOf(allPass(b)), g)
-	if v := get(out, "readiness_verdict"); v != "MEETS_TARGETS_WITH_APPROVED_DEGRADATION" {
-		t.Fatalf("verdict %s", v)
-	}
-}
-
-// 10. Missing gate evidence -> UNKNOWN -> NOT_READY.
-func TestGatePassWithoutEvidence(t *testing.T) {
-	b := base()
-	g := passing()
-	g["gates"].(map[string]any)["G-COVERAGE"] = map[string]any{"status": "PASS"}
-	out := mustScore(t, rubricOf(b), resOf(allPass(b)), g)
-	if get(out, "gates.G-COVERAGE") != "UNKNOWN" || get(out, "readiness_verdict") != "NOT_READY" {
-		t.Fatalf("gate %s verdict %s", get(out, "gates.G-COVERAGE"), get(out, "readiness_verdict"))
-	}
-}
-
-// 11. A rubric cannot lower the policy thresholds or drop a domain without a recorded exclusion.
-func TestWeakenedRubric(t *testing.T) {
-	b := base()
-	low := rubricOf(b)
-	low["policy"] = map[string]any{"global_min": "0"}
-	var four []map[string]any
-	for i, d := range []string{"tests", "architecture", "ux", "operations"} {
-		four = append(four, ctl(i+1, d, 5))
-	}
-	partial := rubricOf(four)
-	partial["domains"] = map[string]any{"tests": 10, "architecture": 10, "ux": 5, "operations": 5}
-	for _, tc := range []struct {
-		r    map[string]any
-		st   map[string]string
-		want string
-	}{{low, allPass(b), "policy may only raise"}, {partial, allPass(four), "needs a weight or a recorded exclusion"}} {
-		if _, err := tryScore(t, tc.r, resOf(tc.st), passing()); err == nil || !strings.Contains(err.Error(), tc.want) {
-			t.Fatalf("want %q, got %v", tc.want, err)
-		}
-	}
-}
-
-func writeJSON(t *testing.T, dir, name string, v any) string {
-	t.Helper()
-	b, err := json.Marshal(v)
-	if err != nil {
-		t.Fatal(err)
-	}
-	p := filepath.Join(dir, name)
-	if err := os.WriteFile(p, b, 0o644); err != nil {
-		t.Fatal(err)
-	}
-	return p
-}
-
-// pinned sets the results' rubric_digest to the digest of the rubric file.
-func pinned(t *testing.T, rubricPath string, results map[string]any) map[string]any {
-	t.Helper()
-	d, err := ovio.SHA256File(rubricPath)
-	if err != nil {
-		t.Fatal(err)
-	}
-	results["rubric_digest"] = d
-	return results
-}
-
-func runCLI(args ...string) (int, string, string) {
-	var o, e bytes.Buffer
-	code := Run(args, &o, &e)
-	return code, o.String(), e.String()
-}
-
-func TestCompare(t *testing.T) {
-	var excl []any
-	for _, d := range doms[2:] {
-		excl = append(excl, map[string]any{"domain": d, "reason": "out of scope", "evidence": []string{"EV-x"}})
-	}
-	r := map[string]any{"schema": "overhaul.rubric/1", "rev": 1, "lanes": []string{"backend"},
-		"domains": map[string]any{"correctness": 50, "security": 50}, "domain_exclusions": excl,
-		"weight_deviations": []any{
-			map[string]any{"domain": "correctness", "reason": "two-domain library", "approval": "AP-1"},
-			map[string]any{"domain": "security", "reason": "two-domain library", "approval": "AP-1"}},
-		"controls": []map[string]any{ctl(1, "correctness", 5), ctl(2, "correctness", 5), ctl(3, "security", 5)}}
-	dir := t.TempDir()
-	rp := writeJSON(t, dir, "r.json", r)
-	code, stdout, stderr := runCLI("compare", "--rubric", rp,
-		"--baseline", writeJSON(t, dir, "a.json", pinned(t, rp, resOf(map[string]string{"C-001": "PASS", "C-002": "UNKNOWN", "C-003": "PASS"}))),
-		"--candidate", writeJSON(t, dir, "b.json", pinned(t, rp, resOf(map[string]string{"C-001": "FAIL", "C-002": "PASS", "C-003": "PASS"}))))
-	if code != 0 {
-		t.Fatalf("exit %d: %s", code, stderr)
-	}
-	out := norm(t, json.RawMessage(stdout))
-	if get(out, "baseline_Q.exact") != "15/2" || get(out, "candidate_Q.exact") != "15/2" {
-		t.Fatalf("Q %v / %v", out["baseline_Q"], out["candidate_Q"])
-	}
-	rows := ovio.List(out["transitions"])
-	if len(rows) != 2 ||
-		get(rows[0], "control") != "C-001" || get(rows[0], "kind") != "regression-or-new-failure" ||
-		get(rows[1], "control") != "C-002" || get(rows[1], "kind") != "evidence-only" {
-		t.Fatalf("transitions %v", rows)
-	}
-}
-
-func TestCLIOutWritesScorecard(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	outPath := filepath.Join(dir, "scorecard.json")
-	rp := writeJSON(t, dir, "r.json", rubricOf(b))
-	code, stdout, stderr := runCLI("--rubric", rp,
-		"--results", writeJSON(t, dir, "s.json", pinned(t, rp, resOf(allPass(b)))),
-		"--gates", writeJSON(t, dir, "g.json", passing()), "--out", outPath)
-	if code != 0 || stdout != "MEETS_TARGETS -> "+outPath+"\n" {
-		t.Fatalf("exit %d stdout %q stderr %q", code, stdout, stderr)
-	}
-	card, err := ovio.LoadObject(outPath)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if get(card, "schema") != "overhaul.scorecard/1" || get(card, "readiness_verdict") != "MEETS_TARGETS" ||
-		len(get(card, "inputs.rubric_sha256")) != 64 {
-		t.Fatalf("scorecard %v", card)
-	}
-}
-
-func TestCLIInvalidInputExits2(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	bad := rubricOf(b)
-	bad["schema"] = "overhaul.rubric/0"
-	rubric := writeJSON(t, dir, "r.json", rubricOf(b))
-	stale := resOf(allPass(b))
-	stale["rubric_digest"] = strings.Repeat("0", 64)
-	for name, args := range map[string][]string{
-		"bad rubric": {"--rubric", writeJSON(t, dir, "bad.json", bad), "--results", writeJSON(t, dir, "s.json", resOf(allPass(b))),
-			"--gates", writeJSON(t, dir, "g.json", passing())},
-		"stale digest": {"--rubric", rubric, "--results", writeJSON(t, dir, "stale.json", stale),
-			"--gates", filepath.Join(dir, "g.json")},
-	} {
-		code, stdout, stderr := runCLI(args...)
-		if code != 2 || stdout != "" || !strings.HasPrefix(stderr, "ERROR: ") {
-			t.Fatalf("%s: exit %d stdout %q stderr %q", name, code, stdout, stderr)
-		}
-	}
-}
-
-// A domain weight off the default table needs a recorded deviation with an approval.
-func TestDomainWeightDeviation(t *testing.T) {
-	b := base()
-	r := rubricOf(b)
-	r["domains"].(map[string]any)["security"] = 30
-	if _, err := tryScore(t, r, resOf(allPass(b)), passing()); err == nil || !strings.Contains(err.Error(), "weight_deviations") {
-		t.Fatalf("undeclared deviation accepted: %v", err)
-	}
-	r["weight_deviations"] = []any{map[string]any{"domain": "security", "reason": "payments service"}}
-	if _, err := tryScore(t, r, resOf(allPass(b)), passing()); err == nil || !strings.Contains(err.Error(), "approval reference") {
-		t.Fatalf("deviation without approval accepted: %v", err)
-	}
-	dup := map[string]any{"domain": "security", "reason": "payments service", "approval": "AP-1"}
-	r["weight_deviations"] = []any{dup, dup}
-	if _, err := tryScore(t, r, resOf(allPass(b)), passing()); err == nil || !strings.Contains(err.Error(), "unknown or repeated") {
-		t.Fatalf("repeated deviation accepted: %v", err)
-	}
-	r["weight_deviations"] = []any{dup}
-	mustScore(t, r, resOf(allPass(b)), passing())
-}
-
-// A performance PASS from static inspection earns nothing.
-func TestStaticPerformancePassIsUnknown(t *testing.T) {
-	b := base()
-	b[3] = maps.Clone(b[3])
-	b[3]["evaluation"] = "static"
-	out := mustScore(t, rubricOf(b), resOf(allPass(b)), passing())
-	if get(out, "status_by_control."+b[3]["id"].(string)) != "UNKNOWN" || get(out, "readiness_verdict") != "NOT_READY" {
-		t.Fatalf("static performance PASS counted: %v", out["status_by_control"])
-	}
-}
-
-// Results without the rubric digest, or compared against another rubric, are invalid.
-func TestRubricDigestRequired(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	rp := writeJSON(t, dir, "r.json", rubricOf(b))
-	other := rubricOf(b)
-	other["rev"] = 2
-	op := writeJSON(t, dir, "other.json", other)
-	g := writeJSON(t, dir, "g.json", passing())
-	for name, args := range map[string][]string{
-		"score without digest": {"--rubric", rp, "--results", writeJSON(t, dir, "s.json", resOf(allPass(b))), "--gates", g},
-		"compare other rubric": {"compare", "--rubric", rp,
-			"--baseline", writeJSON(t, dir, "a.json", pinned(t, rp, resOf(allPass(b)))),
-			"--candidate", writeJSON(t, dir, "c.json", pinned(t, op, resOf(allPass(b))))},
-	} {
-		if code, _, stderr := runCLI(args...); code != 2 || !strings.Contains(stderr, "rubric") {
-			t.Fatalf("%s: exit %d %q", name, code, stderr)
-		}
-	}
-}
-
-// A results document that names no code state is invalid input: exit 2, one
-// diagnostic naming the absent field, and no scorecard written.
-func TestResultsMustIdentifyScoredState(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	rp := writeJSON(t, dir, "r.json", rubricOf(b))
-	g := writeJSON(t, dir, "g.json", passing())
-	for i, tc := range []struct {
-		name   string
-		mutate func(map[string]any)
-		want   string
-	}{
-		{"fingerprint absent", func(r map[string]any) { delete(r, "fingerprint") }, "fingerprint"},
-		{"fingerprint null", func(r map[string]any) { r["fingerprint"] = nil }, "fingerprint"},
-		{"fingerprint empty", func(r map[string]any) { r["fingerprint"] = "" }, "fingerprint"},
-		{"fingerprint placeholder", func(r map[string]any) { r["fingerprint"] = "not-a-sha256" }, "fingerprint"},
-		{"fingerprint one short", func(r map[string]any) { r["fingerprint"] = fp[:len(fp)-1] }, "fingerprint"},
-		{"fingerprint non-hex", func(r map[string]any) { r["fingerprint"] = strings.Repeat("z", 64) }, "fingerprint"},
-		{"subject absent", func(r map[string]any) { delete(r, "subject") }, "subject"},
-		{"subject empty", func(r map[string]any) { r["subject"] = "" }, "subject"},
-		{"subject not a string", func(r map[string]any) { r["subject"] = 12 }, "subject"},
-	} {
-		res := pinned(t, rp, resOf(allPass(b)))
-		tc.mutate(res)
-		sp := writeJSON(t, dir, "s.json", res)
-		card := filepath.Join(dir, fmt.Sprintf("card-%d.json", i))
-		other := "fingerprint"
-		if tc.want == "fingerprint" {
-			other = "subject"
-		}
-		code, stdout, stderr := runCLI("--rubric", rp, "--results", sp, "--gates", g, "--out", card)
-		switch {
-		case code != 2:
-			t.Errorf("%s: exit %d, want 2 (stdout %q stderr %q)", tc.name, code, stdout, stderr)
-		case stdout != "":
-			t.Errorf("%s: published %q", tc.name, stdout)
-		case !strings.HasPrefix(stderr, "ERROR: "):
-			t.Errorf("%s: %q is not a tool diagnostic", tc.name, stderr)
-		case !strings.Contains(stderr, tc.want):
-			t.Errorf("%s: %q does not name %s", tc.name, stderr, tc.want)
-		case strings.Contains(stderr, other):
-			t.Errorf("%s: %q names %s, not %s", tc.name, stderr, other, tc.want)
-		case strings.Contains(stderr, "rubric_digest"):
-			t.Errorf("%s: the digest check fired, not the field check: %q", tc.name, stderr)
-		}
-		if _, err := os.Stat(card); err == nil {
-			t.Errorf("%s: wrote a scorecard for an unidentified code state", tc.name)
-		}
-	}
-}
-
-// The exit 2 above is the absent field, not scoring: identified results score,
-// and a failing control is a scorecard at exit 0 carrying the validated
-// fingerprint verbatim.
-func TestIdentifiedResultsStillScore(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	rp := writeJSON(t, dir, "r.json", rubricOf(b))
-	g := writeJSON(t, dir, "g.json", passing())
-	card := filepath.Join(dir, "card.json")
-	code, stdout, stderr := runCLI("--rubric", rp, "--results", writeJSON(t, dir, "s.json", pinned(t, rp, resOf(allPass(b)))),
-		"--gates", g, "--out", card)
-	if code != 0 || stderr != "" {
-		t.Fatalf("identified results: exit %d stdout %q stderr %q", code, stdout, stderr)
-	}
-	c, err := ovio.LoadObject(card)
-	if err != nil {
-		t.Fatal(err)
-	}
-	if get(c, "fingerprint") != fp || get(c, "subject") != "candidate" || get(c, "readiness_verdict") != "MEETS_TARGETS" {
-		t.Fatalf("scorecard %v", c)
-	}
-	st := allPass(b)
-	st[b[0]["id"].(string)] = "FAIL"
-	code, _, stderr = runCLI("--rubric", rp, "--results", writeJSON(t, dir, "f.json", pinned(t, rp, resOf(st))), "--gates", g)
-	if code != 0 || stderr != "" {
-		t.Fatalf("failing control: exit %d stderr %q", code, stderr)
-	}
-	out := mustScore(t, rubricOf(b), resOf(st), passing())
-	if get(out, "readiness_verdict") != "NOT_READY" || get(out, "fingerprint") != fp {
-		t.Fatalf("failing control: verdict %s fingerprint %s", get(out, "readiness_verdict"), get(out, "fingerprint"))
-	}
-}
-
-// compare scores both arms through the same check, so an arm that identifies no
-// code state is rejected with the arm named.
-func TestCompareArmsRequireScoredState(t *testing.T) {
-	dir := t.TempDir()
-	b := base()
-	rp := writeJSON(t, dir, "r.json", rubricOf(b))
-	arm := func(name string, identified bool) string {
-		res := pinned(t, rp, resOf(allPass(b)))
-		if !identified {
-			delete(res, "fingerprint")
-		}
-		return writeJSON(t, dir, name, res)
-	}
-	for _, tc := range []struct {
-		arm       string
-		baseline  string
-		candidate string
-	}{
-		{"baseline", arm("blank-baseline.json", false), arm("ok-candidate.json", true)},
-		{"candidate", arm("ok-baseline.json", true), arm("blank-candidate.json", false)},
-	} {
-		code, stdout, stderr := runCLI("compare", "--rubric", rp, "--baseline", tc.baseline, "--candidate", tc.candidate)
-		if code != 2 || stdout != "" || !strings.HasPrefix(stderr, "ERROR: "+tc.arm+": ") ||
-			!strings.Contains(stderr, "fingerprint") {
-			t.Fatalf("%s arm: exit %d stdout %q stderr %q", tc.arm, code, stdout, stderr)
-		}
-	}
-}