diff --git a/hydra-gates/scripts/lib/test_gate_crashed_checker_is_not_a_finding.sh b/hydra-gates/scripts/lib/test_gate_crashed_checker_is_not_a_finding.sh index 3bcc45950..84da58d4a 100755 --- a/hydra-gates/scripts/lib/test_gate_crashed_checker_is_not_a_finding.sh +++ b/hydra-gates/scripts/lib/test_gate_crashed_checker_is_not_a_finding.sh @@ -422,7 +422,24 @@ else _bad "control FAILED: no evidence the python3 stub ran — the assertions below prove nothing" fi -for _g in 12:nc-input-labels 15:dashboard-antipattern 16:spec-coverage \ +# Gates 6, 7 and 9 joined this list in .github#330. All three invoked their +# helper as `>> log 2>/dev/null || true` — the gate-20 construct verbatim, +# discarding the message AND the status — and then took the verdict from +# `wc -l` on a log nothing had written to. +# +# MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: +# +# working python3 python3 that exits 1 +# [gate-6] orphan-auth: FAIL [gate-6] orphan-auth: PASS +# [gate-7] no-admin-idor: FAIL [gate-7] no-admin-idor: PASS +# [gate-9] semantic-auth: FAIL [gate-9] semantic-auth: PASS +# +# Three SECURITY gates — dead authorization code, IDOR, and auth-attribute +# mismatch — reporting a tree clean that they had not opened. gate-7 is the +# gate a missing helper had already blinded once, over 11 real unguarded +# endpoints (.github#147); this was the second route to the same false green. +for _g in 6:orphan-auth 7:no-admin-idor 9:semantic-auth \ + 12:nc-input-labels 15:dashboard-antipattern 16:spec-coverage \ 17:redundant-controller 18:notification-dialect 19:e2e-coverage; do _num="${_g%%:*}" _name="${_g#*:}" @@ -492,6 +509,155 @@ for _num in 15 16; do fi done +# --------------------------------------------------------------------------- +# GATES 59-64 — A CRASH MUST NOT WEAR A FINDING COUNT (.github#330) +# +# The mirror image of everything above. These six gates take a NUMERIC EXIT +# PROTOCOL from their helper (0 pass / 1 findings / 3 empty scope / 4 n-a / +# 5 tooling missing) and sent every unrecognised code to `_fail`. A crash also +# exits 1, so it arrived at the findings branch — and because a traceback +# contains no `FAIL` lines, `_count '^FAIL'` returned 0 and the runner then ran +# `[ "${_n}" -eq 0 ] && _n=1`, INVENTING a finding count of one for a check +# that had measured nothing. +# +# MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: +# +# working python3 python3 that exits 1 +# [gate-59] unclosable-gate: PASS FAIL — config gate(s) read but never written +# [gate-62] store-plane: NOT APPLICABLE FAIL — 1 naming or discovery violation(s) +# [gate-63] settings-surface: NOT APPL. FAIL — 1 naming or discovery violation(s) +# +# 62 and 63 blocked a repo that does not even HAVE the subject matter. A gate +# that is wrong in the RED direction burns the same credit as one that is wrong +# in the green direction, and it is the reason a suite gets switched off. +# +# TWO ARMS, as everywhere in this file: the planted true positives must still +# be caught with a working interpreter, or "always skip" would satisfy the +# crash arm. +# --------------------------------------------------------------------------- +_hi_app="${_tmp}/highgates" +mkdir -p "${_hi_app}/lib/AppInfo" "${_hi_app}/lib/Service" "${_hi_app}/src" \ + "${_hi_app}/appinfo" +printf '\n\n hg\n\n' \ + > "${_hi_app}/appinfo/info.xml" +# gate-59: a config key READ but never WRITTEN (the docudesk shape, ADR-076). +cat > "${_hi_app}/lib/Service/SettingsInitializer.php" <<'PHP' +config->getAppValue('hg', 'configuration_version', ''); + if ($v === '') { + $this->importFromApp(); + } + } +} +PHP +# gate-64: an AppHost adoption with no OpenRegister autoload prelude (ADR-040). +cat > "${_hi_app}/lib/AppInfo/Application.php" <<'PHP' + "${_hi_app}/src/manifest.json" <<'JSON' +{"name":"hg","menu":[{"label":"Dashboard","icon":"CarSportOutline","route":"/d"}]} +JSON +( + cd "${_hi_app}" || exit 1 + git init -q . + git add -A + git -c user.email=t@t -c user.name=t commit -qm init +) >/dev/null 2>&1 + +# ARM A — working interpreter. The plants must be caught. +_hilogs_ok="${_tmp}/hilogs-ok"; mkdir -p "${_hilogs_ok}" +_hiout_ok="${_tmp}/highgates-ok.txt" +( + cd "${_hi_app}" || exit 1 + HYDRA_GATE_LOG_DIR="${_hilogs_ok}" bash "${_runner}" . > "${_hiout_ok}" 2>&1 +) +for _g in 59:unclosable-gate 60:icon-vocabulary 64:apphost-autoload-prelude; do + _num="${_g%%:*}"; _name="${_g#*:}" + if grep -qE "^\[gate-${_num}\][^:]*: FAIL" "${_hiout_ok}"; then + _ok "gate-${_num} ${_name}: FAILs its planted true positive with a working interpreter" + else + _v=$(grep -oE "^\[gate-${_num}\] [^:]+: [A-Z]+( [A-Z]+)?( \([a-z]+\))?" "${_hiout_ok}" | head -1 | sed 's/^[^:]*: //') + _bad "gate-${_num} ${_name}: returned '${_v:-none}' for a planted true positive — the crash arm below would prove nothing" + fi +done + +# The count must be MEASURED, not invented. Gates 59 and 64 print no `FAIL`- +# prefixed lines at all, so `_count '^FAIL'` was always 0 and the fallback +# `_n=1` fired on EVERY real failure — a number nobody had counted. +if grep -qE '^\[gate-59\][^:]*: FAIL — [0-9]+ config gate\(s\)' "${_hiout_ok}"; then + _ok "gate-59's failure carries a measured count, not a bare verdict" +else + _bad "gate-59's failure line has no measured count" +fi + +# ARM B — the same tree, interpreter dead. No FAIL, no PASS: a named wiring skip. +_hilogs="${_tmp}/hilogs"; mkdir -p "${_hilogs}" +_hiout="${_tmp}/highgates-crash.txt" +( + cd "${_hi_app}" || exit 1 + PATH="${_fakebin}:${PATH}" HYDRA_GATE_LOG_DIR="${_hilogs}" \ + bash "${_runner}" . > "${_hiout}" 2>&1 +) +for _g in 59:unclosable-gate 60:icon-vocabulary 61:listener-work-placement \ + 62:store-plane 63:settings-surface 64:apphost-autoload-prelude; do + _num="${_g%%:*}"; _name="${_g#*:}" + _v=$(grep -oE "^\[gate-${_num}\] [^:]+: [A-Z]+( [A-Z]+)?( \([a-z]+\))?" "${_hiout}" | head -1 | sed 's/^[^:]*: //') + case "${_v}" in + "SKIPPED (wiring)") + _ok "gate-${_num} ${_name}: SKIPPED (wiring) when its checker cannot run" + ;; + FAIL) + _bad "gate-${_num} ${_name}: FAIL from a crashed checker — an environment failure rendered as findings about the source (#233/#245/#330)" + ;; + PASS) + _bad "gate-${_num} ${_name}: PASS over a checker that never executed" + ;; + *) + _bad "gate-${_num} ${_name}: verdict is '${_v:-none emitted}' — expected SKIPPED (wiring)" + ;; + esac +done + +# --------------------------------------------------------------------------- +# THE CONSTRUCT ITSELF — `2>/dev/null || true` on a helper invocation. +# +# Every specific assertion above is a statement about the gates that exist +# today. This one is about the SHAPE, so the next gate to land with it fails +# here on its first run rather than after a fleet-wide false green. +# +# `2>/dev/null` on a helper discards the reason it died; `|| true` discards the +# fact that it died. Together they are why gate-20 had never fired in any repo +# in its entire existence: its grep pattern began with `->`, grep parsed that as +# options and exited 2, and both halves of the evidence went in the bin. +# --------------------------------------------------------------------------- +_offenders=$(grep -nE '(python3|node|npx|bash)[^|]*2>/dev/null[[:space:]]*\|\|[[:space:]]*true' "${_scripts}/run-hydra-gates.sh" \ + | grep -vE '^[0-9]+:[[:space:]]*#' || true) +# Multi-line invocations put the redirect on a continuation line, so also look +# for a `2>/dev/null || true` whose PREVIOUS line ends in a backslash after an +# interpreter name. +_offenders2=$(awk ' + /(python3|node |npx |bash )/ && /\\$/ { prev=NR; next } + /2>\/dev\/null[[:space:]]*\|\|[[:space:]]*true/ && NR == prev+1 { print NR": "$0 } +' "${_scripts}/run-hydra-gates.sh") +if [ -z "${_offenders}${_offenders2}" ]; then + _ok "no helper in run-hydra-gates.sh is invoked with '2>/dev/null || true' (the gate-20 construct)" +else + _bad "a helper is still invoked with '2>/dev/null || true' — message and status both discarded:" + printf '%s\n%s\n' "${_offenders}" "${_offenders2}" | sed '/^$/d' | sed 's/^/ /' +fi + echo if [ "${_failures}" -eq 0 ]; then echo "test_gate_crashed_checker_is_not_a_finding.sh: ALL PASS" diff --git a/hydra-gates/scripts/run-hydra-gates.sh b/hydra-gates/scripts/run-hydra-gates.sh index 109e544ee..7f2c6867a 100755 --- a/hydra-gates/scripts/run-hydra-gates.sh +++ b/hydra-gates/scripts/run-hydra-gates.sh @@ -572,6 +572,47 @@ _count() { printf '%s' "${_n}" } +# _helper_finished — did the checker reach its +# own summary line? +# +# WHY THIS EXISTS (.github#330) — THE MIRROR IMAGE OF THE gate-20 DEFECT +# --------------------------------------------------------------------- +# Gates 59-64 use a NUMERIC EXIT PROTOCOL from their helper: 0 pass, 1 findings, +# 3 empty scope, 4 not applicable, 5 tooling missing. The runner switched on +# those codes and sent everything unrecognised to a `_fail` — so a CRASH, which +# also exits 1, arrived at the findings branch. And because the crash wrote a +# traceback rather than `FAIL` lines, `_count '^FAIL'` came back 0 and the next +# line read `[ "${_n}" -eq 0 ] && _n=1`: the runner INVENTED a finding count of +# one for a check that had measured nothing. +# +# MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: +# +# working python3 python3 that exits 1 +# [gate-59] unclosable-gate: PASS [gate-59] ... FAIL — config gate(s) read but never written +# [gate-62] store-plane: NOT APPL. [gate-62] ... FAIL — 1 naming or discovery violation(s) +# [gate-63] settings-surface: NOT A [gate-63] ... FAIL — 1 naming or discovery violation(s) +# +# Two of those repos do not even HAVE the subject matter the gate blocked them +# on. This is .github#233/#245's "an environment failure rendered as findings +# about the source" — the same disease as gate-20's silent green, wearing red +# instead. A blind gate and a lying gate are both gates you cannot read. +# +# THE EXIT BYTE CANNOT SEPARATE THEM, because 1 means both "I found things" and +# "I died". What CAN separate them is that every one of these helpers prints a +# TERMINAL SUMMARY LINE — `checked N manifest(s): M failure(s)`, +# `N unclosable gate(s).`, `apphost-autoload-prelude: OK` — on every path that +# reaches a verdict, and a process that died before its own summary never +# printed it. That is the same evidence gates 15/16/17 take from their +# `# count=` marker (.github#271), read from the summary these helpers already +# emit rather than from a marker they would have to grow. +# +# The failure direction is deliberate: no marker ⇒ `_skip ... wiring`, which +# still fails a --require-full-coverage run. This converts a fabricated finding +# into an honest "I could not look", never into a pass. +_helper_finished() { + grep -qE "$2" "$1" 2>/dev/null +} + # --------------------------------------------------------------------------- # ADR-040 AppHost adoption — shared between gate-5 (route-auth) and gate-14 # (route-reachability). @@ -2051,8 +2092,36 @@ if [ "${#_oa_files[@]}" -gt 0 ]; then _oa_lib_dir="${SCRIPT_DIR}/lib" fi if [ -f "${_oa_lib_dir}/check_orphan_auth.py" ]; then + # A CHECKER THAT COULD NOT RUN MUST NOT JUDGE THE CODE (.github#330 — + # the #245/#233/#276 rule, applied to the three security gates it had + # not reached). + # + # This was `>> log 2>/dev/null || true`: the helper's message AND its + # exit status both discarded, with the verdict then taken from + # `wc -l` on a log nothing had written to — i.e. PASS. That is the + # gate-20 mechanism verbatim, and gate-20 had NEVER fired in any repo + # in its entire existence because of it. + # + # MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: + # working python3 [gate-6] orphan-auth: FAIL — 1 orphan method(s) + # python3 exit 1 [gate-6] orphan-auth: PASS + # Same tree, same files in scope. The green was over nothing. + # + # The helper ALWAYS returns 0 when it runs, by design (#209 — the + # count goes to stdout, never into the exit byte; see its `main()`). + # So a non-zero exit here is unambiguously a crash and can never be a + # finding count. stderr goes to a .err file rather than /dev/null so + # the reason it died is recoverable by the reader. + _oa_err="${_oa_log}.err" + : > "${_oa_err}" python3 "${_oa_lib_dir}/check_orphan_auth.py" "${_oa_files[@]}" \ - >> "${_oa_log}" 2>/dev/null || true + >> "${_oa_log}" 2>"${_oa_err}" + _oa_rc=$? + if [ "${_oa_rc}" -ne 0 ]; then + _oa_ran=0 + _oa_why=$(head -3 "${_oa_err}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 6 "orphan-auth" wiring "check_orphan_auth.py exited ${_oa_rc} — ${#_oa_files[@]} PHP file(s) were in scope and NONE were judged; orphaned (defined-but-never-called) authorization methods are UNVERIFIED by this run. This helper always exits 0 when it runs, so this is a crash, not a finding count. Checker output: ${_oa_why:-}. See ${_oa_err}." + fi else _oa_ran=0 _skip 6 "orphan-auth" wiring "check_orphan_auth.py not found at ${_oa_lib_dir} — ${#_oa_files[@]} PHP file(s) were in scope and NONE were inspected; orphaned (defined-but-never-called) authorization methods are UNVERIFIED by this run." @@ -2129,8 +2198,31 @@ if [ "${#_idor_files[@]}" -gt 0 ]; then _gate_lib_dir="${SCRIPT_DIR}/lib" fi if [ -f "${_gate_lib_dir}/check_no_admin_idor.py" ]; then + # A CHECKER THAT COULD NOT RUN MUST NOT JUDGE THE CODE (.github#330). + # Same repair, same reason, same measurement as gate-6 above — see the + # note there. This gate's green history is the one that matters most: + # a missing helper already made it report PASS over 11 real unguarded + # IDOR endpoints once (.github#147), and `2>/dev/null || true` left the + # SECOND route to that same false green — a helper that is present but + # cannot run — wide open. + # + # MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: + # working python3 [gate-7] no-admin-idor: FAIL — 1 method(s) + # python3 exit 1 [gate-7] no-admin-idor: PASS + # + # check_no_admin_idor.py ends `return 0 # exit 0 always — caller + # counts printed lines`, so a non-zero byte here is a crash and never + # a count. + _idor_err="${_idor_log}.err" + : > "${_idor_err}" python3 "${_gate_lib_dir}/check_no_admin_idor.py" "${_idor_files[@]}" \ - >> "${_idor_log}" 2>/dev/null || true + >> "${_idor_log}" 2>"${_idor_err}" + _idor_rc=$? + if [ "${_idor_rc}" -ne 0 ]; then + _idor_ran=0 + _idor_why=$(head -3 "${_idor_err}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 7 "no-admin-idor" wiring "check_no_admin_idor.py exited ${_idor_rc} — ${#_idor_files[@]} controller file(s) were in scope and NONE were judged; unguarded #[NoAdminRequired] endpoints (IDOR, OWASP A01:2021) are UNVERIFIED by this run. This helper always exits 0 when it runs, so this is a crash, not a finding count. Checker output: ${_idor_why:-}. See ${_idor_err}." + fi else _idor_ran=0 _skip 7 "no-admin-idor" wiring "check_no_admin_idor.py not found at ${_gate_lib_dir} — ${#_idor_files[@]} controller file(s) were in scope and NONE were inspected; unguarded #[NoAdminRequired] endpoints (IDOR, OWASP A01:2021) are UNVERIFIED by this run." @@ -2262,8 +2354,26 @@ if [ "${#_sem_files[@]}" -gt 0 ]; then fi done if [ -n "${_sem_helper}" ]; then + # A CHECKER THAT COULD NOT RUN MUST NOT JUDGE THE CODE (.github#330). + # Same repair as gates 6 and 7 above. + # + # MEASURED 2026-08-10 at package 71e01bf, one fixture, two runs: + # working python3 [gate-9] semantic-auth: FAIL — 1 mismatch(es) + # python3 exit 1 [gate-9] semantic-auth: PASS + # + # check_semantic_auth.py ends `return 0 # exit 0 always — caller + # counts printed lines`, so a non-zero byte here is a crash and never + # a count. + _sem_err="${_sem_log}.err" + : > "${_sem_err}" python3 "${_sem_helper}" "${_sem_files[@]}" \ - >> "${_sem_log}" 2>/dev/null || true + >> "${_sem_log}" 2>"${_sem_err}" + _sem_rc=$? + if [ "${_sem_rc}" -ne 0 ]; then + _sem_ran=0 + _sem_why=$(head -3 "${_sem_err}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 9 "semantic-auth" wiring "check_semantic_auth.py exited ${_sem_rc} — ${#_sem_files[@]} controller file(s) were in scope and NONE were judged; auth-attribute-vs-body semantic mismatches are UNVERIFIED by this run. This helper always exits 0 when it runs, so this is a crash, not a finding count. Checker output: ${_sem_why:-}. See ${_sem_err}." + fi else _sem_ran=0 _skip 9 "semantic-auth" wiring "check_semantic_auth.py not found near ${SCRIPT_DIR} — ${#_sem_files[@]} controller file(s) were in scope and NONE were inspected; auth-attribute-vs-body semantic mismatches are UNVERIFIED by this run." @@ -7597,8 +7707,18 @@ if [ -d lib ] && [ "${_ucg_scoped_out}" -eq 0 ]; then set +e if [ "${_ucg_rc}" -eq 0 ]; then _pass 59 "unclosable-gate" + elif ! _helper_finished "${_ucg_log}" '^([0-9]+ unclosable gate\(s\)\.|unclosable-gate: OK)$'; then + # A CRASH IS NOT A FINDING (.github#330) — see _helper_finished. This + # branch used to be the only alternative to PASS, so a python3 that + # exited 1 was reported as "config gate(s) read but never written". + _ucg_why=$(head -3 "${_ucg_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 59 "unclosable-gate" wiring "check_unclosable_gate.py exited ${_ucg_rc} without printing its terminal summary line, so NO PHP composition root was judged and read-but-never-written config gates (ADR-076 rule 3) are UNVERIFIED by this run. Checker output: ${_ucg_why:-}. See ${_ucg_log}." else - _fail 59 "unclosable-gate" "config gate(s) read but never written — guarded setup runs on every request (ADR-076 rule 3); see ${_ucg_log}" + # The helper does not prefix its findings with `FAIL`, so the count + # comes off its own terminal summary line rather than being invented. + _ucg_n=$(grep -oE '^[0-9]+ unclosable gate\(s\)\.$' "${_ucg_log}" 2>/dev/null | grep -oE '^[0-9]+' | tail -1) + case "${_ucg_n}" in ''|*[!0-9]*) _ucg_n=1 ;; esac + _fail 59 "unclosable-gate" "${_ucg_n} config gate(s) read but never written — guarded setup runs on every request (ADR-076 rule 3); see ${_ucg_log}" fi elif [ ! -d lib ]; then _skip 59 "unclosable-gate" na "no lib/ in this repo — there is no PHP composition root in which a config gate could be read." @@ -7670,6 +7790,10 @@ if [ -f src/manifest.json ]; then # FAILs when node_modules was absent, then — once those were # guarded — a silent PASS over a rule that never executed (#233). _skip 60 "icon-vocabulary" wiring "vue-material-design-icons is not installed, so icon names could NOT be verified to exist upstream — an invented MDI name would render blank and this run would not have caught it (ADR-077 rule 1). Every other icon rule passed. Run npm ci to restore full coverage. See ${_iv_log}." + elif ! _helper_finished "${_iv_log}" '^(checked [0-9]+ manifest\(s\):|no manifest in scope)'; then + # A CRASH IS NOT A FINDING (.github#330) — see _helper_finished. + _iv_why=$(head -3 "${_iv_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 60 "icon-vocabulary" wiring "check_icon_vocabulary.py exited ${_iv_rc} without printing its terminal 'checked N manifest(s)' summary, so NO icon was judged and the ADR-077 vocabulary is UNVERIFIED by this run. Checker output: ${_iv_why:-}. See ${_iv_log}." else _iv_n=$(_count '^FAIL' "${_iv_log}") [ "${_iv_n}" -eq 0 ] && _iv_n=1 @@ -7739,6 +7863,10 @@ if [ -d lib/AppInfo ]; then _skip 61 "listener-work-placement" na "the diff against '${BASE_REF}' put every post-event registration out of scope, so NONE were inspected. Diff-scoped out under ADR-020 — the fleet's 149-registration backlog is a work-list, not a reason to block an unrelated PR. This gate runs on the next PR that touches a listener registration. See ${_lwp_log}." elif [ "${_lwp_rc}" -eq 4 ]; then _skip 61 "listener-work-placement" na "this repo registers no post-object-event listener, so there is no work to place on or off the write path (ADR-078). See ${_lwp_log}." + elif ! _helper_finished "${_lwp_log}" '^checked [0-9]+ (post-event )?registration\(s\)'; then + # A CRASH IS NOT A FINDING (.github#330) — see _helper_finished. + _lwp_why=$(head -3 "${_lwp_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 61 "listener-work-placement" wiring "check_listener_placement.py exited ${_lwp_rc} without printing its terminal 'checked N ... registration(s)' summary, so NO post-object-event listener was judged and ADR-078 work placement is UNVERIFIED by this run. Checker output: ${_lwp_why:-}. See ${_lwp_log}." else _lwp_n=$(_count '^FAIL' "${_lwp_log}") [ "${_lwp_n}" -eq 0 ] && _lwp_n=1 @@ -7776,6 +7904,16 @@ elif [ "${_sp_rc}" -eq 3 ]; then _skip 62 "store-plane" na "the diff against '${BASE_REF}' touched no manifest and no menu-layout, so NO manifest was inspected. Diff-scoped out under ADR-020 — not a gap: the manifests in this repo are unchanged from the base branch, so this PR introduces no store-plane naming or discovery decision (ADR-080) to judge. This gate runs on the next PR that touches a manifest. See ${_sp_log}." elif [ "${_sp_rc}" -eq 4 ]; then _skip 62 "store-plane" na "no src/manifest.json — a Tier-0 app declares no store plane for ADR-080 to constrain." +elif ! _helper_finished "${_sp_log}" '^(checked [0-9]+ manifest\(s\):|FAIL )'; then + # A CRASH IS NOT A FINDING (.github#330) — see _helper_finished. MEASURED: + # on a fixture with no src/manifest.json at all, where the honest verdict + # is NOT APPLICABLE, a python3 that exits 1 made this gate report + # `FAIL — 1 store/templates/catalogue naming or discovery violation(s)`. + # The `FAIL ` alternative in the marker keeps the helper's early + # unparseable-JSON return — a real finding printed before the summary — + # on the failing side where it belongs. + _sp_why=$(head -3 "${_sp_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 62 "store-plane" wiring "check_store_and_settings_surface.py --gate store exited ${_sp_rc} without printing its terminal 'checked N manifest(s)' summary or any FAIL line, so NO manifest was judged and ADR-080 store-plane naming/discovery is UNVERIFIED by this run. Checker output: ${_sp_why:-}. See ${_sp_log}." else _sp_n=$(_count '^FAIL' "${_sp_log}") [ "${_sp_n}" -eq 0 ] && _sp_n=1 @@ -7830,6 +7968,10 @@ elif [ "${_ss_rc}" -eq 3 ]; then _skip 63 "settings-surface" na "the diff against '${BASE_REF}' touched no manifest and no menu-layout, so NO manifest was inspected. Diff-scoped out under ADR-020 — not a gap. This gate adjudicates ADR-079 placement as declared in src/manifest.json, src/manifest.d/ and src/menu-layout.json, and nothing else. ${_ss_extra} It runs on the next PR that touches one of those. See ${_ss_log}." elif [ "${_ss_rc}" -eq 4 ]; then _skip 63 "settings-surface" na "no src/manifest.json — a Tier-0 app declares no settings surface for ADR-079 to place." +elif ! _helper_finished "${_ss_log}" '^(checked [0-9]+ manifest\(s\):|FAIL )'; then + # A CRASH IS NOT A FINDING (.github#330) — see gate 62 above, same helper. + _ss_why=$(head -3 "${_ss_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 63 "settings-surface" wiring "check_store_and_settings_surface.py --gate settings exited ${_ss_rc} without printing its terminal 'checked N manifest(s)' summary or any FAIL line, so NO manifest was judged and ADR-079 settings placement is UNVERIFIED by this run. Checker output: ${_ss_why:-}. See ${_ss_log}." else _ss_n=$(_count '^FAIL' "${_ss_log}") [ "${_ss_n}" -eq 0 ] && _ss_n=1 @@ -7893,9 +8035,15 @@ if [ -d lib/AppInfo ]; then grep -E '^NOTE' "${_aap_log}" || true if [ "${_aap_rc}" -eq 0 ]; then _pass 64 "apphost-autoload-prelude" + elif ! _helper_finished "${_aap_log}" '^([0-9]+ AppHost adoption\(s\) without the autoload prelude\.|apphost-autoload-prelude: OK)$'; then + # A CRASH IS NOT A FINDING (.github#330) — see _helper_finished. + _aap_why=$(head -3 "${_aap_log}" 2>/dev/null | tr '\n' ' ' | cut -c1-200) + _skip 64 "apphost-autoload-prelude" wiring "check_apphost_autoload_prelude.py exited ${_aap_rc} without printing its terminal summary line, so NO register() body was judged and ADR-040 autoload preludes are UNVERIFIED by this run. Checker output: ${_aap_why:-}. See ${_aap_log}." else - _aap_n=$(_count '^FAIL' "${_aap_log}") - [ "${_aap_n}" -eq 0 ] && _aap_n=1 + # The helper does not prefix its findings with `FAIL`, so the count + # comes off its own terminal summary line rather than being invented. + _aap_n=$(grep -oE '^[0-9]+ AppHost adoption\(s\) without the autoload prelude\.$' "${_aap_log}" 2>/dev/null | grep -oE '^[0-9]+' | tail -1) + case "${_aap_n}" in ''|*[!0-9]*) _aap_n=1 ;; esac _fail 64 "apphost-autoload-prelude" "${_aap_n} AppHost adoption(s) with no OpenRegister autoload prelude (ADR-040); see ${_aap_log}" fi else