diff --git a/CHANGELOG.md b/CHANGELOG.md index e0c9250..3ddb045 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -14,6 +14,7 @@ so entries say what changed, cite the issue, and stop. - `offsite` — protect claimed issues whose PR lives in another repository from the claim-reclaim clock (#68). - `issueflow-reconcile` — nudge once when an `offsite` flag outlives every visible cross-referenced PR (#69). - Ruling doctrine — define every human-owned trigger, the fixed escalation shape, and the 0–24h builder-to-triage ladder (#72). +- The sweep observes the escalation contract: a malformed escalation is named field-by-field, and the ladder's 12h/24h rungs each draw one comment to the flag-setter — comment-only, per-episode, both surfaces (#73). ## 0.1.0 — 2026-07-22 diff --git a/lib/ruling.sh b/lib/ruling.sh index a2a2882..119a5cf 100644 --- a/lib/ruling.sh +++ b/lib/ruling.sh @@ -1,11 +1,13 @@ #!/usr/bin/env bash -# lib/ruling.sh — the `needs-ruling` sweep invariants (issue #52, epic #50). +# lib/ruling.sh — the `needs-ruling` sweep invariants (issues #52 and #73, +# epic #50). # -# Both reconcilers source this file: the bare-flag check and the 7-day nudge -# are ONE implementation serving both surfaces — two copies of a 7-day rule -# is how the family got here in the first place (#50). Pure decisions sit -# above the divider (facts in, verdict out); the one impure orchestrator -# below talks to gh and posts through the sourcing script's run()/log(). +# Both reconcilers source this file: the bare-flag check, the escalation +# shape check, the ladder's rung comments and the 7-day nudge are ONE +# implementation serving both surfaces — two copies of a 7-day rule is how +# the family got here in the first place (#50). Pure decisions sit above +# the divider (facts in, verdict out); the one impure orchestrator below +# talks to gh and posts through the sourcing script's run()/log(). # # Standing rules this file lives under: # - The machine never sets or clears `needs-ruling` (#50 D9). Nothing here @@ -30,9 +32,22 @@ RULING_NUDGE_AFTER=$((7 * 24 * 3600)) # The escalation back-window: a comment this many seconds before the # `labeled` event still accompanies it. RULING_BARE_WINDOW=$((15 * 60)) -# The idempotency marker for the bare-flag comment. The nudge deliberately +# The ladder's rungs (#50 D13, via #72's doctrine): moments past the current +# episode's `labeled` event. Two timers only — the 24h rung's comment names +# triage's past-24h authority too, so there is no third timer to keep honest. +RULING_RUNG12_AT=$((12 * 3600)) +RULING_RUNG24_AT=$((24 * 3600)) +# The escalation contract's four field labels (#50 D12). Literal strings — +# the template in BUILDER.md ("the ruling ask") fixes them so this machinery +# can check for them; presence is all that is ever checked (#50 D4). +RULING_SHAPE_FIELDS=(Options: Recommend: Blocked: Default:) +# The idempotency markers, one per comment kind, each scoped to the current +# `labeled` episode by ruling_bare_comment_needed. The nudge deliberately # has NO marker — see ruling_nudge_decision. RULING_BARE_MARKER='' +RULING_SHAPE_MARKER='' +RULING_RUNG12_MARKER='' +RULING_RUNG24_MARKER='' # --------------------------------------------------------------------------- # Pure decisions. Facts in (args/stdin), verdict out. No gh, no clock. @@ -74,6 +89,8 @@ ruling_bare_comment_needed() { # $1 labeled epoch, $2 newest marked-comment epoc # → POST | SKIP. Scoped to the CURRENT labeled event: a marked comment # older than the event belongs to an earlier flag episode, so a genuine # re-flag is re-checked while a 15-minute cron never repeats itself. + # Marker-agnostic — the caller tracks a newest epoch PER marker (#73), so + # this one comparison scopes every marked write (bare, shape, both rungs). local labeled="$1" marked="${2:-}" if [ -n "$marked" ] && [ "$marked" -gt "$labeled" ]; then echo SKIP @@ -82,6 +99,59 @@ ruling_bare_comment_needed() { # $1 labeled epoch, $2 newest marked-comment epoc fi } +ruling_shape_decision() { # escalation body on stdin → SHAPED | MALFORMED + # Presence only (#50 D4): that `Recommend:` exists is checkable, that the + # recommendation is any good is not — no counting options, no parsing the + # prose. Line-anchored, allowing leading whitespace and Markdown bold + # (`**Options:**` is how the live escalations write them): the labels + # appearing only mid-sentence is not the template. The `🧭 needs-ruling` + # header line is deliberately unchecked — it is prose, and an emoji grep + # on an LC_ALL=C runner is a portability trap for zero enforcement value. + local body field missing="" + body="$(cat)" + for field in "${RULING_SHAPE_FIELDS[@]}"; do + grep -Eq "^[[:space:]]*(\*\*)?$field" <<<"$body" || missing="$missing $field" + done + if [ -z "$missing" ]; then echo SHAPED; else echo "MALFORMED$missing"; fi +} + +ruling_deadline_decision() { # $1 now, $2 the current episode's labeled epoch → RUNG0 | RUNG12 | RUNG24 + # The ladder anchors to the `labeled` event, never to activity (#50 D14: + # an active back-and-forth still climbs it; only the separate 7-day nudge + # resets). "At 12h" means AT: the boundary starts the rung — the rung is a + # moment whose duty exists the moment it strikes, unlike the strictly-past + # nudge horizon. RUNG24 covers "past 24h" too: the 24h comment names both + # the builder's rung and triage's past-24h authority, so there is no + # fourth timer to keep honest. + local age=$(($1 - $2)) + if [ "$age" -ge "$RULING_RUNG24_AT" ]; then + echo RUNG24 + elif [ "$age" -ge "$RULING_RUNG12_AT" ]; then + echo RUNG12 + else + echo RUNG0 + fi +} + +ruling_default_decision() { # escalation body on stdin → DEADLINE | HARDBLOCK | UNPARSEABLE + # Parsed only to *describe* the item in the rung comments, never to gate a + # rung (#50 D14: the rungs apply whatever `Default:` says). Mechanical or + # nothing: an ISO-8601 UTC timestamp anywhere on the `Default:` line is + # the deadline, the literal word `none` is a hard block, anything else is + # reported as unparseable rather than guessed at. Only the `Default:` line + # is read — a timestamp elsewhere in the body is somebody's prose. + local line ts + line="$(grep -E '^[[:space:]]*(\*\*)?Default:' | head -n1)" + ts="$(grep -oE '[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}(:[0-9]{2})?Z' <<<"$line" | head -n1)" + if [ -n "$ts" ]; then + echo "DEADLINE $ts" + elif grep -qE '(^|[^[:alnum:]])none([^[:alnum:]]|$)' <<<"$line"; then + echo HARDBLOCK + else + echo UNPARSEABLE + fi +} + ruling_nudge_decision() { # $1 now, $2 last real-activity epoch → NUDGE | KEEP # Real activity only — comments, reviews, commits, never label churn, or # the sweep would reset its own clock. The nudge needs NO marker: the @@ -99,21 +169,29 @@ ruling_newest_flag() { # "loginiso8601" lines on stdin → the newest line sort -t $'\t' -k2 | tail -n1 } -ruling_escalation_url() { # $1 setter, $2 labeled epoch; "login epoch url" lines on stdin - # → the url of the EARLIEST in-window comment by the setter, or nothing. +ruling_escalation_row() { # $1 setter, $2 labeled epoch; "login epoch url [b64]" lines on stdin + # → "url b64" of the EARLIEST in-window comment by the setter, or nothing. # Earliest, because the natural shape is escalation-then-flag: the first # qualifying comment is the escalation itself, later ones are follow-ups. - local setter="$1" labeled="$2" login epoch url best_epoch="" best_url="" - while read -r login epoch url; do + # The body rides along base64-encoded (#73's shape check reads it); rows + # without the column still resolve, with an empty body. + local setter="$1" labeled="$2" login epoch url b64 best_epoch="" best="" + while read -r login epoch url b64; do [ -n "$login" ] || continue [ "$login" = "$setter" ] || continue ruling_accompanies "$epoch" "$labeled" || continue if [ -z "$best_epoch" ] || [ "$epoch" -lt "$best_epoch" ]; then best_epoch="$epoch" - best_url="$url" + best="$url ${b64:-}" fi done - [ -z "$best_url" ] || printf '%s\n' "$best_url" + [ -z "$best" ] || printf '%s\n' "$best" +} + +ruling_escalation_url() { # same contract, url column only — the nudge's link + local row + row="$(ruling_escalation_row "$@")" + [ -z "$row" ] || printf '%s\n' "${row%% *}" } # --------------------------------------------------------------------------- @@ -148,33 +226,49 @@ reconcile_ruling() { # $1 item number, $2 last real-activity epoch, $3 now labeled_at="${newest##*$'\t'}" labeled_epoch="$(date -d "$labeled_at" +%s)" + # The body travels as jq's @base64 — bodies carry newlines and tabs, and + # the whole file is line-oriented, so the row format stays TSV and the + # body is decoded at its points of use (#73). Do not switch rows to JSON. local comments if ! comments="$(gh api --paginate "repos/$REPO/issues/$n/comments" \ --jq '.[] | [.user.login, .created_at, .html_url, - (if ((.body // "") | contains("")) - then "marked" else "plain" end)] | @tsv' 2>/dev/null)"; then + ((.body // "") | @base64)] | @tsv' 2>/dev/null)"; then log "#$n: ruling comments unreadable — no verdict invented this pass" return 0 fi - # One pass over the comments builds every fact the decisions consume: - # who commented when (for the bare verdict), the newest marked comment - # (for idempotency), and the "login epoch url" rows the link lookup reads. - local login at url kind epoch authored="" rows="" marked="" - while IFS=$'\t' read -r login at url kind; do + # One pass over the comments builds every fact the decisions consume: who + # commented when (for the bare verdict), the newest marked comment PER + # MARKER (each write is idempotent per episode — one pass, not one pass + # per marker), and the "login epoch url b64" rows the escalation lookup + # reads. A body that fails to decode counts as unmarked — for idempotency + # that risks a repeat, never an invented verdict. + local login at url b64 body epoch authored="" rows="" + local marked_bare="" marked_shape="" marked_rung12="" marked_rung24="" + while IFS=$'\t' read -r login at url b64; do [ -n "$login" ] || continue epoch="$(date -d "$at" +%s)" authored="$authored$login $epoch"$'\n' - rows="$rows$login $epoch $url"$'\n' - if [ "$kind" = marked ]; then - if [ -z "$marked" ] || [ "$epoch" -gt "$marked" ]; then marked="$epoch"; fi - fi + rows="$rows$login $epoch $url ${b64:-}"$'\n' + body="$(base64 -d <<<"${b64:-}" 2>/dev/null)" || body="" + case "$body" in *"$RULING_BARE_MARKER"*) + if [ -z "$marked_bare" ] || [ "$epoch" -gt "$marked_bare" ]; then marked_bare="$epoch"; fi ;; + esac + case "$body" in *"$RULING_SHAPE_MARKER"*) + if [ -z "$marked_shape" ] || [ "$epoch" -gt "$marked_shape" ]; then marked_shape="$epoch"; fi ;; + esac + case "$body" in *"$RULING_RUNG12_MARKER"*) + if [ -z "$marked_rung12" ] || [ "$epoch" -gt "$marked_rung12" ]; then marked_rung12="$epoch"; fi ;; + esac + case "$body" in *"$RULING_RUNG24_MARKER"*) + if [ -z "$marked_rung24" ] || [ "$epoch" -gt "$marked_rung24" ]; then marked_rung24="$epoch"; fi ;; + esac done <<<"$comments" # ---- the bare-flag check (#50 D4, mechanical proxy) ---- - if [ "$(ruling_bare_decision "$setter" "$labeled_epoch" <<<"$authored")" = BARE ] \ - && [ "$(ruling_bare_comment_needed "$labeled_epoch" "$marked")" = POST ]; then - run gh issue comment "$n" -R "$REPO" --body "$RULING_BARE_MARKER + if [ "$(ruling_bare_decision "$setter" "$labeled_epoch" <<<"$authored")" = BARE ]; then + if [ "$(ruling_bare_comment_needed "$labeled_epoch" "$marked_bare")" = POST ]; then + run gh issue comment "$n" -R "$REPO" --body "$RULING_BARE_MARKER The ruling flag on this item was set by @$setter with no accompanying escalation comment. Setting it requires the escalation contract — the **question**, the **options**, and a **recommendation** — posted by the @@ -183,7 +277,88 @@ after ([LABELS.md](https://github.com/heavy-duty/ceremony/blob/main/LABELS.md) carries the flag-setter's obligations; heavy-duty/ceremony#50 D4). The label stays — this machine never removes an escalation on the strength of a timestamp heuristic — but the contract is still owed." >/dev/null - log "#$n: ruling flag is bare — commented (the label is never removed)" + log "#$n: ruling flag is bare — commented (the label is never removed)" + fi + # Bare stops here (#73): there is no shape to check when there is no + # escalation comment, and a rung comment beside the bare comment would be + # two comments about the same omission — noise. The 7-day nudge below is + # deliberately untouched by this exclusion; it predates the ladder and + # already words the bare case itself. + else + # ---- the shape check (#50 D12): the contract's four field labels ---- + local esc_row esc_url esc_body shape + esc_row="$(ruling_escalation_row "$setter" "$labeled_epoch" <<<"$rows")" + esc_url="${esc_row%% *}" + if ! esc_body="$(base64 -d <<<"${esc_row#* }" 2>/dev/null)"; then + # An undecodable body must not become "malformed" — an unreadable fact + # never invents a verdict. The rungs read the same body, so they wait + # for a readable pass too. + log "#$n: escalation body unreadable — no verdict invented this pass" + else + shape="$(ruling_shape_decision <<<"$esc_body")" + if [ "$shape" != SHAPED ] \ + && [ "$(ruling_bare_comment_needed "$labeled_epoch" "$marked_shape")" = POST ]; then + local missing="${shape#MALFORMED }" + run gh issue comment "$n" -R "$REPO" --body "$RULING_SHAPE_MARKER +@$setter — the [escalation comment]($esc_url) accompanying this ruling flag +is missing required field labels: **$missing**. The contract's shape is +fixed because this machinery checks for it (heavy-duty/ceremony#50 D12): +four line-anchored field labels — \`Options:\`, \`Recommend:\`, \`Blocked:\`, +\`Default:\` — per the canonical template in +[BUILDER.md — the ruling ask](https://github.com/heavy-duty/ceremony/blob/main/BUILDER.md#the-ruling-ask). +Presence is all that is checked; the machine never judges the prose +(heavy-duty/ceremony#50 D4). The label stays — the shape is owed, not +enforced." >/dev/null + log "#$n: escalation malformed (missing:$missing) — commented (the shape is owed, not enforced)" + fi + + # ---- the ladder's rungs (#50 D13–D14), observed never decided ---- + # Each rung's comment fires once per episode, AT its moment: a rung + # whose moment passed unobserved (the sweep was down through 12h–24h) + # is not paged after the fact — the later rung's comment carries the + # whole remaining duty. + local rung state described + rung="$(ruling_deadline_decision "$now" "$labeled_epoch")" + state="$(ruling_default_decision <<<"$esc_body")" + case "$state" in + DEADLINE\ *) described="a stated default deadline of \`${state#DEADLINE }\`" ;; + HARDBLOCK) described="\`Default: none\` — a hard block; no default ever fires" ;; + *) described="a missing or unparseable \`Default:\` line — reported as-is, never guessed at (and the contract's own rule is that unsure is a hard block)" ;; + esac + if [ "$rung" = RUNG12 ] \ + && [ "$(ruling_bare_comment_needed "$labeled_epoch" "$marked_rung12")" = POST ]; then + run gh issue comment "$n" -R "$REPO" --body "$RULING_RUNG12_MARKER +@$setter — this ruling is 12 hours past its \`labeled\` event: the ladder's +12h rung ([BUILDER.md — the ruling ask](https://github.com/heavy-duty/ceremony/blob/main/BUILDER.md#the-ruling-ask), +heavy-duty/ceremony#50 D13). Mechanically read, the escalation carries +$described. + +The rung's duty is the flag-setter's: re-read the \`Default:\` against +everything that has landed since the flag went up — does it still hold, and +has reasonable doubt appeared? A stale default does not fire, and new doubt +makes it a hard block. The rungs run on the \`labeled\` clock and do not +reset on activity; this comment fires once per flag episode." >/dev/null + log "#$n: ruling at the 12h rung — commented (the setter re-reads the default)" + fi + if [ "$rung" = RUNG24 ] \ + && [ "$(ruling_bare_comment_needed "$labeled_epoch" "$marked_rung24")" = POST ]; then + run gh issue comment "$n" -R "$REPO" --body "$RULING_RUNG24_MARKER +@$setter — this ruling is 24 hours past its \`labeled\` event: the ladder's +24h rung ([BUILDER.md — the ruling ask](https://github.com/heavy-duty/ceremony/blob/main/BUILDER.md#the-ruling-ask), +heavy-duty/ceremony#50 D13). Mechanically read, the escalation carries +$described. + +At 24h the builder proceeds regardless, **as a PR**: pick an option and +state in the PR body which way you went and what doubt remains. Nothing +merges by this — the human still gates the merge. Past 24h the choice is +triage's to make: triage picks the option, records it as a decision, and +remains accountable; the operator may overturn it at merge. The rungs run on +the \`labeled\` clock and do not reset on activity; this comment fires once +per flag episode and covers everything past 24h — there is no further +timer." >/dev/null + log "#$n: ruling at the 24h rung — commented (the builder proceeds as a PR; past 24h is triage's)" + fi + fi fi # ---- the 7-day nudge (#50 D10) ---- diff --git a/test/issueflow-reconcile.test.sh b/test/issueflow-reconcile.test.sh index 52e65a5..6b0d708 100644 --- a/test/issueflow-reconcile.test.sh +++ b/test/issueflow-reconcile.test.sh @@ -253,8 +253,17 @@ cfix() { printf '%s/repos_owner_repo_issues_%s_comments.json' "$TMP" "$1"; } jq -n --arg l "$(iso_at $((INOW - 10 * 86400)))" \ '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$l}, {"event":"assigned","created_at":$l}]' >"$(tfix 21)" +# The escalation is conforming and the rung markers are pre-seeded — by 10 +# days in both rungs fired long ago (#73), so this probe observes the nudge +# wiring alone; shape and rung behavior have their own probes in +# test/ruling.test.sh. jq -n --arg at "$(iso_at $((INOW - 10 * 86400 - 60)))" \ - '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc21","body":"question, options, recommendation"}]' \ + --arg b $'Options: A — x B — y\nRecommend: A, because x.\nBlocked: z\nDefault: none — hard block' \ + --arg r12 "$(iso_at $((INOW - 10 * 86400 + 13 * 3600)))" \ + --arg r24 "$(iso_at $((INOW - 10 * 86400 + 25 * 3600)))" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc21","body":$b}, + {"user":{"login":"sweep-bot"},"created_at":$r12,"html_url":"https://x/r12","body":"\nrung"}, + {"user":{"login":"sweep-bot"},"created_at":$r24,"html_url":"https://x/r24","body":"\nrung"}]' \ >"$(cfix 21)" exempt="$(issue_probe 21 $'claimed\nneeds-ruling')" check "a 10-day-quiet claim under a ruling is not reclaimed" 1 "" \ diff --git a/test/labels-reconcile.test.sh b/test/labels-reconcile.test.sh index fa89f25..6f5765e 100755 --- a/test/labels-reconcile.test.sh +++ b/test/labels-reconcile.test.sh @@ -557,12 +557,22 @@ ruling_sweep_probe() { # $1 = the PR's labels → reconcile_pr's log lines ) } -# The flag went up 8 days ago with its escalation posted seconds earlier. -jq -n --arg at "$(iso_at $((RNOW - 8 * 86400)))" \ +# The flag went up 10 days ago with its escalation posted seconds earlier; +# the newest activity is the reviews at 8 days. The escalation is conforming +# and the rung markers are pre-seeded — by now both rungs fired long ago +# (#73), older than the reviews so the quiet window still reads 8 days — +# and this probe observes the nudge wiring alone; shape and rung behavior +# have their own probes in test/ruling.test.sh. +jq -n --arg at "$(iso_at $((RNOW - 10 * 86400)))" \ '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ >"$RTMP/repos_owner_repo_issues_77_timeline.json" -jq -n --arg at "$(iso_at $((RNOW - 8 * 86400 - 60)))" \ - '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc77","body":"question, options, recommendation"}]' \ +jq -n --arg at "$(iso_at $((RNOW - 10 * 86400 - 60)))" \ + --arg b $'Options: A — x B — y\nRecommend: A, because x.\nBlocked: z\nDefault: none — hard block' \ + --arg r12 "$(iso_at $((RNOW - 10 * 86400 + 13 * 3600)))" \ + --arg r24 "$(iso_at $((RNOW - 10 * 86400 + 25 * 3600)))" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc77","body":$b}, + {"user":{"login":"sweep-bot"},"created_at":$r12,"html_url":"https://x/r12","body":"\nrung"}, + {"user":{"login":"sweep-bot"},"created_at":$r24,"html_url":"https://x/r24","body":"\nrung"}]' \ >"$RTMP/repos_owner_repo_issues_77_comments.json" wired="$(ruling_sweep_probe "needs-ruling")" diff --git a/test/ruling.test.sh b/test/ruling.test.sh index fa3a3ed..2eaffb7 100644 --- a/test/ruling.test.sh +++ b/test/ruling.test.sh @@ -51,12 +51,63 @@ check "exactly 7 days holds — strictly past the horizon, like the stale sweep" ruling_nudge_decision "$NOW" $((NOW - 7 * 86400)) check "fresh activity holds" 0 "KEEP" ruling_nudge_decision "$NOW" $((NOW - 60)) +# -- the ladder's rungs (#50 D13): anchored to the labeled event ------------ + +check "11h59m is rung 0" 0 "RUNG0" ruling_deadline_decision "$NOW" $((NOW - 12 * 3600 + 60)) +check "exactly 12h starts the rung — at means at, unlike the nudge horizon" 0 "RUNG12" \ + ruling_deadline_decision "$NOW" $((NOW - 12 * 3600)) +check "12h01m is rung 12" 0 "RUNG12" ruling_deadline_decision "$NOW" $((NOW - 12 * 3600 - 60)) +check "23h59m is still rung 12" 0 "RUNG12" ruling_deadline_decision "$NOW" $((NOW - 24 * 3600 + 60)) +check "exactly 24h starts the last rung" 0 "RUNG24" ruling_deadline_decision "$NOW" $((NOW - 24 * 3600)) +check "25h is rung 24 — past-24h has no fourth timer" 0 "RUNG24" \ + ruling_deadline_decision "$NOW" $((NOW - 25 * 3600)) + +# -- the Default: line, parsed for wording only (#50 D14) ------------------- + +check "a timed default parses to its deadline" 0 "DEADLINE 2026-07-23T21:00Z" \ + ruling_default_decision <<<$'Blocked: x\nDefault: A at 2026-07-23T21:00Z if no ruling' +check "a bold default line with seconds parses" 0 "DEADLINE 2026-07-23T21:00:30Z" \ + ruling_default_decision <<<$'**Default:** A at 2026-07-23T21:00:30Z if quiet' +check "none is a hard block" 0 "HARDBLOCK" ruling_default_decision <<<$'Default: none — hard block' +check "prose is unparseable, never guessed" 0 "UNPARSEABLE" \ + ruling_default_decision <<<$'Default: when it feels right' +check "a missing default line is unparseable" 0 "UNPARSEABLE" \ + ruling_default_decision <<<$'Options: A\nRecommend: A.' +check "a timestamp off the default line is somebody's prose" 0 "UNPARSEABLE" \ + ruling_default_decision <<<$'the deadline 2026-07-23T21:00Z came up above\nDefault: soonish' + rows="$(printf 'setter %s https://x/first\nsetter %s https://x/late\nbystander %s https://x/other\nsetter %s https://x/early-out\n' \ "$((L - 300))" "$((L + 600))" "$((L - 60))" "$((L - 5000))")" check "the nudge links the earliest in-window escalation by the setter" 0 "https://x/first" \ ruling_escalation_url setter "$L" <<<"$rows" check "no qualifying escalation yields no link" 0 "" \ ruling_escalation_url setter "$L" <<<"bystander $((L - 60)) https://x/other" +check "the url survives the base64 body column" 0 "https://x/first" \ + ruling_escalation_url setter "$L" <<<"setter $((L - 300)) https://x/first $(printf 'Options: A' | base64)" +check "the row carries the body for the shape check" 0 "https://x/first $(printf 'Options: A' | base64)" \ + ruling_escalation_row setter "$L" <<<"setter $((L - 300)) https://x/first $(printf 'Options: A' | base64)" + +# -- the escalation comment's shape (#50 D12): presence only, line-anchored -- + +TPL=$'🧭 needs-ruling — fixture decision\nOptions: A — on B — off\nRecommend: A, because the drill says so.\nBlocked: the fixture stops; everything else continues\nDefault: A at 2026-07-23T21:00Z if no ruling' +TPL_BOLD=$'🧭 needs-ruling — fixture decision\n**Options:** A — on B — off\n**Recommend:** A, because the drill says so.\n**Blocked:** the fixture stops\n**Default:** none — hard block' + +check "all four field labels present is shaped" 0 "SHAPED" ruling_shape_decision <<<"$TPL" +check "bold field labels are shaped — the live escalations write them bold" 0 "SHAPED" \ + ruling_shape_decision <<<"$TPL_BOLD" +check "labels inside a details fold are shaped — line-anchored, not fold-aware" 0 "SHAPED" \ + ruling_shape_decision <<<$'
fold\nOptions: A — x B — y\nRecommend: A.\nBlocked: nothing\nDefault: none\n
' +check "leading whitespace is tolerated" 0 "SHAPED" \ + ruling_shape_decision <<<$' Options: A B\n Recommend: A.\n Blocked: x\n Default: none' +check "one missing label is named" 0 "MALFORMED Recommend:" \ + ruling_shape_decision <<<$'Options: A — x B — y\nBlocked: z\nDefault: none — hard block' +check "two missing labels are both named" 0 "MALFORMED Recommend: Default:" \ + ruling_shape_decision <<<$'Options: A — x B — y\nBlocked: z' +check "labels only mid-sentence are malformed — line-anchoring is the rule" 0 \ + "MALFORMED Options: Recommend: Blocked: Default:" \ + ruling_shape_decision <<<$'we should Options: A or B, and I Recommend: A; Blocked: no; Default: none' +check "an empty body is missing everything" 0 "MALFORMED Options: Recommend: Blocked: Default:" \ + ruling_shape_decision ' "$TMP/posted-9" check "the bare comment names the missing contract" 0 "" \ grep -q 'question' "$TMP/posted-9" +check "a bare flag draws no shape or rung comment — bare stops here" 1 "" \ + grep -qE 'needs-ruling-(shape|rung)' "$TMP/posted-9" -# -- accompanied flag: silence ---------------------------------------------- +# -- accompanied, conforming (bold, the live spelling): silence -------------- jq -n --arg at "$(iso "$T")" \ '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ >"$(timeline_file 10)" -jq -n --arg at "$(iso "$((T - 840))")" \ - '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc","body":"question, options, recommendation"}]' \ +jq -n --arg at "$(iso "$((T - 840))")" --arg b "$TPL_BOLD" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc","body":$b}]' \ >"$(comments_file 10)" reconcile_ruling 10 "$T" "$NOW" >/dev/null -check "an accompanied flag posts nothing" 0 "0" posts 10 +check "an accompanied conforming flag posts nothing" 0 "0" posts 10 # -- re-flag: judged on its own escalation, marker scoped per event ---------- T1=$((NOW - 86400)) T2=$((NOW - 7200)) @@ -148,12 +201,18 @@ reconcile_ruling 11 "$T2" "$NOW" >/dev/null check "a re-flag is re-checked against its own escalation" 0 "1" posts 11 # -- nudge: fires past 7 quiet days, links the escalation, resets itself ----- +# The escalation is conforming and the rung markers are pre-seeded (by 8 days +# in, both rungs fired long ago — the realistic board), so the nudge's +# behavior is observed alone. T0=$((NOW - 8 * 86400)) jq -n --arg at "$(iso "$T0")" \ '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ >"$(timeline_file 12)" -jq -n --arg at "$(iso "$((T0 - 60))")" \ - '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc12","body":"question, options, recommendation"}]' \ +jq -n --arg at "$(iso "$((T0 - 60))")" --arg b "$TPL" \ + --arg r12 "$(iso "$((T0 + 13 * 3600))")" --arg r24 "$(iso "$((T0 + 25 * 3600))")" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc12","body":$b}, + {"user":{"login":"sweep-bot"},"created_at":$r12,"html_url":"https://x/r12","body":"\nrung"}, + {"user":{"login":"sweep-bot"},"created_at":$r24,"html_url":"https://x/r24","body":"\nrung"}]' \ >"$(comments_file 12)" check "8 quiet days nudge" 0 "ruling nudge" reconcile_ruling 12 "$T0" "$NOW" check "one nudge posted" 0 "1" posts 12 @@ -170,21 +229,143 @@ check "the posted nudge is now the newest activity" 0 "" \ reconcile_ruling 12 "$(date -d "$newest_at" +%s)" "$NOW" >/dev/null check "a sweep right after the nudge holds its silence" 0 "1" posts 12 -# -- 6 quiet days: silence --------------------------------------------------- -jq -n --arg at "$(iso "$((NOW - 6 * 86400))")" \ +# -- 6 quiet days: silence. The flag itself is fresh (label churn is not +# activity, so a quiet item can be freshly flagged): rung 0, no nudge. ------ +jq -n --arg at "$(iso "$((NOW - 3600))")" \ '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ >"$(timeline_file 13)" -jq -n --arg at "$(iso "$((NOW - 6 * 86400))")" \ - '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc13","body":"question, options, recommendation"}]' \ +jq -n --arg at "$(iso "$((NOW - 3660))")" --arg b "$TPL" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc13","body":$b}]' \ >"$(comments_file 13)" reconcile_ruling 13 $((NOW - 6 * 86400)) "$NOW" >/dev/null -check "6 quiet days do not nudge" 0 "0" posts 13 +check "6 quiet days do not nudge, and a fresh flag sits on rung 0" 0 "0" posts 13 # -- unreadable timeline: nothing happens ------------------------------------ check "an unreadable timeline invents no verdict" 0 "timeline unreadable" \ reconcile_ruling 14 "$T" "$NOW" check "...and posts nothing" 0 "0" posts 14 +# -- malformed escalation, rung 0: the shape comment, exactly once ----------- +T15=$((NOW - 3600)) +jq -n --arg at "$(iso "$T15")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 15)" +jq -n --arg at "$(iso "$((T15 - 60))")" \ + --arg b $'Options: A — x B — y\nBlocked: z\nDefault: none — hard block' \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc15","body":$b}]' \ + >"$(comments_file 15)" +check "a malformed escalation is commented on" 0 "escalation malformed" \ + reconcile_ruling 15 "$T15" "$NOW" +reconcile_ruling 15 "$T15" "$NOW" >/dev/null +check "one shape comment across two sweeps" 0 "1" posts 15 +check "the shape comment names exactly the missing label" 0 "" \ + grep -qF 'missing required field labels: **Recommend:**' "$TMP/posted-15" +check "the shape comment links the escalation" 0 "" grep -qF 'https://x/esc15' "$TMP/posted-15" +check "the shape comment quotes the template location" 0 "" \ + grep -qF 'BUILDER.md#the-ruling-ask' "$TMP/posted-15" +check "no bare comment beside the shape comment" 1 "" \ + grep -qF "$RULING_BARE_MARKER" "$TMP/posted-15" +check "the shape comment carries its marker" 0 "" \ + grep -qF "$RULING_SHAPE_MARKER" "$TMP/posted-15" + +# -- 13h in, activity minutes old: the 12h rung fires anyway (D14) ----------- +T16=$((NOW - 13 * 3600)) +jq -n --arg at "$(iso "$T16")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 16)" +jq -n --arg at "$(iso "$((T16 - 60))")" --arg b "$TPL_BOLD" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc16","body":$b}]' \ + >"$(comments_file 16)" +check "the 12h rung fires despite recent activity — rungs never reset" 0 "12h rung" \ + reconcile_ruling 16 $((NOW - 60)) "$NOW" +reconcile_ruling 16 $((NOW - 60)) "$NOW" >/dev/null +check "one 12h rung comment, once per episode" 0 "1" posts 16 +check "the rung comment is addressed to the flag-setter" 0 "" grep -qF '@setter' "$TMP/posted-16" +check "the rung comment names the hard block" 0 "" grep -qF 'Default: none' "$TMP/posted-16" +check "no nudge rode along — activity is recent and the nudge does reset" 1 "" \ + grep -qF '@danmt' "$TMP/posted-16" +check "the rung comment carries its marker" 0 "" \ + grep -qF "$RULING_RUNG12_MARKER" "$TMP/posted-16" + +# -- the ladder walked on the cron: 12h rung at 13h, 24h rung at 25h --------- +L17=$((NOW - 25 * 3600)) +jq -n --arg at "$(iso "$L17")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 17)" +jq -n --arg at "$(iso "$((L17 - 60))")" --arg b "$TPL" \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc17","body":$b}]' \ + >"$(comments_file 17)" +reconcile_ruling 17 "$L17" $((L17 + 13 * 3600)) >/dev/null +check "a sweep at 13h posts the 12h rung" 0 "1" posts 17 +check "the rung comment names the stated deadline" 0 "" \ + grep -qF '2026-07-23T21:00Z' "$TMP/posted-17" +check "a sweep at 25h posts the 24h rung — one comment per rung" 0 "24h rung" \ + reconcile_ruling 17 "$L17" "$NOW" +check "two rung comments total" 0 "2" posts 17 +check "the 24h comment names triage's past-24h authority" 0 "" \ + grep -qF 'triage picks the option' "$TMP/posted-17" +reconcile_ruling 17 "$L17" "$NOW" >/dev/null +check "...and never a third within the episode" 0 "2" posts 17 + +# -- first observed past 24h: the missed 12h moment is not paged after the +# fact — the 24h comment carries the whole remaining duty --------------------- +T18=$((NOW - 25 * 3600)) +jq -n --arg at "$(iso "$T18")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 18)" +jq -n --arg at "$(iso "$((T18 - 60))")" \ + --arg b $'Options: A — x B — y\nRecommend: A, because x.\nBlocked: z\nDefault: when it feels right' \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc18","body":$b}]' \ + >"$(comments_file 18)" +reconcile_ruling 18 "$T18" "$NOW" >/dev/null +check "a rung first observed past 24h pages once, not retroactively" 0 "1" posts 18 +check "only the 24h comment fired" 1 "" grep -qF "$RULING_RUNG12_MARKER" "$TMP/posted-18" +check "the unparseable default is reported, not guessed" 0 "" \ + grep -qF 'unparseable' "$TMP/posted-18" + +# -- a re-flag climbs its own ladder: old rung markers belong to the old +# episode ------------------------------------------------------------------- +TA=$((NOW - 3 * 86400)) TB=$((NOW - 13 * 3600)) +jq -n --arg ta "$(iso "$TA")" --arg tb "$(iso "$TB")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$ta}, + {"event":"unlabeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$tb}, + {"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$tb}]' \ + >"$(timeline_file 19)" +jq -n --arg esc "$(iso "$((TB - 60))")" --arg b "$TPL" \ + --arg r12 "$(iso "$((TA + 13 * 3600))")" --arg r24 "$(iso "$((TA + 25 * 3600))")" \ + '[{"user":{"login":"sweep-bot"},"created_at":$r12,"html_url":"https://x/r12","body":"\nold episode"}, + {"user":{"login":"sweep-bot"},"created_at":$r24,"html_url":"https://x/r24","body":"\nold episode"}, + {"user":{"login":"setter"},"created_at":$esc,"html_url":"https://x/esc19","body":$b}]' \ + >"$(comments_file 19)" +reconcile_ruling 19 "$TB" "$NOW" >/dev/null +check "a re-flag climbs its own ladder — old rung markers do not stick" 0 "1" posts 19 +check "the re-flag's rung comment is the 12h rung" 0 "" \ + grep -qF "$RULING_RUNG12_MARKER" "$TMP/posted-19" + +# -- unreadable comment list: nothing happens -------------------------------- +jq -n --arg at "$(iso "$T")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 20)" +check "an unreadable comment list invents no verdict" 0 "comments unreadable" \ + reconcile_ruling 20 "$T" "$NOW" +check "...and it posts nothing" 0 "0" posts 20 + +# -- malformed AND on a rung: the shape is owed and the ladder still climbs -- +T21=$((NOW - 13 * 3600)) +jq -n --arg at "$(iso "$T21")" \ + '[{"event":"labeled","label":{"name":"needs-ruling"},"actor":{"login":"setter"},"created_at":$at}]' \ + >"$(timeline_file 21)" +jq -n --arg at "$(iso "$((T21 - 60))")" \ + --arg b $'Options: A — x B — y\nRecommend: A, because x.\nBlocked: z' \ + '[{"user":{"login":"setter"},"created_at":$at,"html_url":"https://x/esc21","body":$b}]' \ + >"$(comments_file 21)" +reconcile_ruling 21 "$T21" "$NOW" >/dev/null +check "a malformed escalation still climbs the ladder — two comments" 0 "2" posts 21 +check "the shape half fired" 0 "" grep -qF "$RULING_SHAPE_MARKER" "$TMP/posted-21" +check "the rung half fired" 0 "" grep -qF "$RULING_RUNG12_MARKER" "$TMP/posted-21" +reconcile_ruling 21 "$T21" "$NOW" >/dev/null +check "both halves are once-per-episode" 0 "2" posts 21 + # -- across every scenario above: not one label write ------------------------ check "the ruling sweep never wrote a label" 1 "" test -f "$TMP/edits"