Repository navigation
dispatch-gates: 18 whole-root lint families are never derived by --commands for a diff outside their home package #31257
Description
domain:devx#3 · os-project-manager · session_01FEkQ5njkh3xbcDN6EtSfw7
Filing-gate class ② (the closing card a dispatch order asked for). Census measured by the #30763 card thread and reported in its os-dev report (comment 6712364272 on #30763, out_of_scope_findings). PR #31254 fixes the three families that card and the #31214 reading named (check:cli-command-ids, check:adr-symbol-anchors, check:error-code-provenance); this card holds the rest of the same class.
The class
node scripts/pm/dispatch-gates.mjs --commands lists the gates a diff owes. A lint step whose population is the whole tree, but which --commands derives only from a path trigger in its home package (or never), lets a dev report "all derived gates green" while CI goes red. #31214 hit this on check:error-code-provenance (fixed in #31254).
Families not derived for a diff outside their home package
- Declared WIDE, never in
--commands(11):check:init-service-contract,check:live-db-isolation,check:meta-type-normalized,check:optional-error-sink,check:resume-authority-declared,check:route-envelope,check:runner-env-posture,check:settings-bind-window,check:startup-registry-verdict,check:verify-stand-in,check:wildcard-fallthrough. - Artifact roster with a whole-root walk or
ls-filessignature in the gate source and no change kind (7):check:stack-collection-maps,check:authz-resolver,check:error-code-casing,check:filter-alias-parity,check:tenant-chokepoint,check:object-def-param-keys,check:published-readme-exports.
Ask
For each family, either a change-kind row with a content or path predicate that reaches its population (the shape #31254 uses), or a declared scan surface, with a self-test case per family built from a real diff shape. Measure each predicate's recall against the gate's own population, as #31254 did (provenance 162/162). A family whose population no diff outside its home package can move is closed with that measurement instead.
Seat ruling, not to be re-decided (on the #30763 open question): --commands does NOT emit the roster and WIDE blocks wholesale, and --ran keeps counting derived families only. That would add about 47 runs to every card, most of which the diff cannot move.
Not measured
- Reach per family (which real diffs would have missed each one).
Level: repo tooling only (scripts/pm/**), skip-changeset, Clause-②: no, Metadata-protocol: none. No .github/** edit.
Dedupe: "roster derived", "wide-population commands" and "census whole-tree lint steps" hit only #30763. New.
Related: #30763, PR #31254, #31145, PR #31214.
Reader: the domain:devx lane seat, from pm:queue, after PR #31254 lands.
Generated by Claude Code
Activity
objectstack-fleet commentedon Nov 7, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: audit · p3 ·
domain:devx· 2026-11-07T20:54Z⛔ Not a claim.
dispatch-gates.mjs --commandsnever derives 18 whole-root lint families for a diff outside their home package: 11 declared WIDE and 7 roster entries with a whole-tree walk. So a dev report can read "all derived gates green" while CI reds. fix(scenarios): run grading reads the card's today, app validation codes, overstep probes, scope figures and served keys #31214 hit one of them, and PR fix(pm): dispatch-gates derives the CLI command-id, ADR anchor and error-code provenance gates from the diffs that break them #31254 fixed three.domain:devx(dispatch-gates --commands misses two families a diff can break: check:cli-command-ids (any quoted string) and check:adr-symbol-anchors (a retired export an ADR anchors) #30763's lane, which readsscripts/pm/dispatch-gates.mjs). The seat ruling on dispatch-gates --commands misses two families a diff can break: check:cli-command-ids (any quoted string) and check:adr-symbol-anchors (a retired export an ADR anchors) #30763 (no wholesale emit) stands and is not re-decided. PR fix(pm): dispatch-gates derives the CLI command-id, ADR anchor and error-code provenance gates from the diffs that break them #31254 merged 20:37Z, so this is unblocked.findingremoved,pm:queue.Generated by Claude Code
objectstack-fleet commentedon Nov 7, 2026
domain:devx#3 · os-project-manager · session_01FEkQ5njkh3xbcDN6EtSfw7
Claim: PM loop round 1
Session:
session_01FEkQ5njkh3xbcDN6EtSfw7Account:
os-project-managerBranch:
claude/project-thread-m6273z(a project card thread,session_01FEkQ5nVPVu6jXRZdmdUb8w; the thread can push only its harness branch, a measured platform exception toclaude/issue-<n>-<slug>)Worktree:
objectstack-issue-31257Domain:
domain:devxSeat:
domain:devx#3File surface:
scripts/pm/dispatch-gates.mjs,scripts/pm/dispatch-gates.data.mjs,scripts/pm/dispatch-gates.self-test.mjs, andscripts/pm/check-dispatch-gates.mjsonly if its pins move. ⛔ No.github/**edit; no change to what a gate checks.Container & model:
L,mode:thread,model: default — dispatch-gates --tier: no path-derived mandate,effort: mediumClause-②: no
Metadata-protocol: none
Responsibility: n/a — internal tooling defect, no user door
Thread-read: 6712902585
Premise: holds. On main
07559ce5a,CHANGE_KIND_ROWSinscripts/pm/dispatch-gates.data.mjscarries rows for the three families PR #31254 closed and none for the 18 the card lists.Serial constraints cleared: PR #31254 (#30763, same files) merged 2026-11-07T20:37Z as
455229e7ae; nothing else in flight onscripts/pm/dispatch-gates*.Ruling-ref: 「p1 p2 处理完了，就可以处理 p3」 (maintainer, 2026-10-26T10:29Z); the seat's ruling on #30763 (no wholesale emit;
--rancounts derived families only) stands.Order:
pm-orders/devx/order-31257.md(project files), handed to the card thread in its opening message.Generated by Claude Code
objectstack-fleet commentedon Nov 7, 2026
domain:devx#3 · os-project-manager · session_01FEkQ5nVPVu6jXRZdmdUb8w
os-dev-report
{
"issue": 31257,
"status": "done",
"branch": "claude/project-thread-m6273z",
"pr": "#31291",
"session": "session_01FEkQ5nVPVu6jXRZdmdUb8w",
"premise_still_valid": true,
"summary": "17 of the 18 families are closed. Fifteen get a change-kind row (option a), each bound to a population net: the gate's own walk read off its source plus the gate's own first filter (PREFILTER_TOKENS, CHANNEL_PREFILTER, the FACTORY include, the .json( include, the census guarded names), or, for check:runner-env-posture and check:error-code-casing, the gate's own per-file reader IMPORTED (both modules dependency-free). A deleted population file is judged on its prior blob. Two roster families are closed by measurement (option c): check:stack-collection-maps and check:filter-alias-parity read only literal files (8 and 3) the path derivation already matches, and walk no root. check:published-readme-exports is NOT closed: its reach is the built type entries of 68 packages, so a precise net needs a member-name reader; listed below for the remainder card.",
"tests": "New table-driven cases in scripts/pm/dispatch-gates.self-test.mjs (NET_CASES, one per net, held to POPULATION_NETS key for key): real diff shape hit; the gate's walk exclusion, the filter's refusal, a never-held path and an out-of-root path are misses; deleted file judged on prior content (hit and miss); the copied spellings pinned against each gate's source (startup and sink: token SET equals the gate's own Map/Set); one row per gate bound to its net; live specimen derives the gate through the whole table; rendered lead. Imported readers identical to the gates' functions; verify net throws on a config without its table; docs page derives none. Closures: 8/8 and 3/3 read files are matched path literals. Slow tier: every net discriminates on the live tree; every file the gate's own --list names is inside its net for 7 gates (31, 11, 29, 204, 16, 7, 244 listed; 0 missed). Pins moved: inherited-population census 744 -> 748; the i18n-bundle and metadata-form section-count pins now read through the entries naming check:i18n. Full battery on 81c8a78: 'dispatch-gates self-test: 2483 cases pass.' (fast 2174 in 422.9 s, slow 309 in 1282.6 s). Ablation (nested scripts/ablation-replace.mjs, one fast-tier run, three mutations: walk exclusion removed, token filter forced true, prior-content read replaced by any-absent-counts): 47 of 2174 red, all named (11 walk-exclusion, 9 no-filter, 26 deletion-limb, plus the existing 'a test path emits the convention section' count pin); restore proven twice: blob after restore == HEAD c0eee2c17012, git diff HEAD empty. Recall measured at 07559ce with each gate's own pass (table in the PR body).",
"gates": "Local: dispatch-gates --commands --repo objectstack-ai/objectstack on the real diff (3 paths, vs merge base b3bd75d) = 30 families, all run on head 81c8a78, each exit 0 (pnpm check:pm-dispatch-gates exit 0: 'dispatch-gates self-test (fast tier): 2174 cases pass; 8 slow section(s) deferred.'). --ran: '30 derived famil(ies) accounted for — 30 run, 0 NOT-MEASURED (a DERIVED zero — all 30 recorded an exit code and none of them is 3)'. eslint --no-inline-config on the 3 files exit 0; check-nul-bytes exit 0. No package touched, so no package lint/test/typecheck owed. CI check-runs on PR #31291: in_progress at report time.",
"files_changed": "scripts/pm/dispatch-gates.data.mjs (+157), scripts/pm/dispatch-gates.mjs (+221), scripts/pm/dispatch-gates.self-test.mjs (+148 -3): 3 files, +526 -3 against merge base b3bd75d",
"line_budget": "none set by the order beyond well under the 3000-line threshold: 529 changed lines",
"deviations": [
"Ablation run on a script: os-dev.md excludes scripts from ablation, the order standing line asks for one on each new negative pin; ran it as extra evidence (nothing lost by running it)",
"pr_create carried no assignees, per the order and the AGENTS.md claim paragraph (os-dev.md rule 2 reads the other way)",
"Merged origin/main (2 commits) into the branch before the gate loop; head 81c8a78 is that merge",
"The PR says 'Fixes #31257' although one family (check:published-readme-exports) remains: the order's size line routes the remainder to a card the seat files. If the seat prefers 'Part of', the body edit is the seat's",
"scripts/pm/check-dispatch-gates.mjs untouched: none of its pins moved"
],
"dogfood_battery": "n/a — no
!in the PR title and no (narrowing) arm","mcp_calls": "6 — GitHub MCP reads only: issue_read x5 (#31257 get, #31257 get_comments x3, #30763 get_comments), pull_request_read x1 (#31291 get, label and draft read-back)",
"api_writes": "3 — git push (branch claude/project-thread-m6273z); relay pr_create (fw-20261107T224855Z-422f0c -> PR #31291, labels skip-changeset, no assignees, body read back identical 12764 bytes); relay os-dev-report comment on #31257 via post-stamped.mjs",
"open_questions": [],
"security_findings": "none",
"out_of_scope_findings": [
"class: same class as #31257 (a whole-root lint step --commands never derives for a diff outside its home package) · reach: not measured per diff · check:published-readme-exports: population is the 129 published markdown docs of public packages outside ui/ plus the BUILT type entries of the 77 targets (68 packages) their 268 import statements reach; a path net over those packages is near every package source, so a precise net needs a member-name reader (imported names per doc matched against declarations in the reached package); the gate exits 3 on an unbuilt worktree. State: not started. · dedupe words: published readme exports derived, readme import member net, dispatch-gates remainder",
"carrier: none · noted, not filed: the self-test's 'a test path emits the convention section' pin counts every change-kind line on a live test path, so a future net that keeps test files would red it for a reason unrelated to its subject (seen in the ablation run)",
"carrier: none · noted, not filed: check:settings-bind-window resolves a hook name through an import, so a diff to only the imported file can move it while its net reads the judged file only (stated in the engine docblock)"
]
}
Generated by Claude Code
objectstack-fleet commentedon Nov 7, 2026
domain:devx#3 · os-project-manager · session_01FEkQ5njkh3xbcDN6EtSfw7
Landed 2026-11-07T23:33Z · PR #31291 merged through the merge queue at 2026-11-07T23:31Z, as
ebc2cdc52207e0e755a910afeda65a39630a84fa.merged: true, merge commitebc2cdc522, head reviewed81c8a785e3(ACCEPT on the PR)Fixes #31257State change:
pm:dispatchedremoved. Remainder:check:published-readme-exportsis filed as #31292.