objectstack build / validate print at most 50 author-time warnings and drop the rest with no notice

The author-time warning printer in @objectstack/cli 17.1.0 emits a fixed 50 detailed entries and then stops. The summary line still counts every warning, so the two numbers disagree — and nothing in the output says the list was cut. A repo past 50 warnings therefore carries diagnostics that are counted but never shown, and an author who reads the output and sees their file is clean has read a truncated list.

Measured

Two objectstack build runs on objectstack-ai/hotcrm, before and after a five-warning fix (hotcrm#1248), same tree otherwise:

run          detailed entries printed     summary line
before       50                           ⚠ 80 author-time warning(s) — see above
after        50                           ⚠ 75 author-time warning(s) — see above

Both runs produced output of exactly 184 lines and exactly 52 ⚠ lines. Removing five warnings did not shorten the printed list — it made room, and five warnings that had never been printed before appeared in the second run:

⚠ view "views[12]" · form: … absolute colSpan 2 …
    rule: absolute-colspan-discouraged
ℹ flow "opportunity_approval_on_create" · node "manager_review": …
    approval-approvers-may-resolve-empty          (×4)

Those five were present in the tree on both runs. They were simply past the cut on the first one. objectstack validate prints the same truncated set.

Why it matters

The truncation is silent — there is no …and 30 more line — so the output is not merely incomplete, it is indistinguishable from complete. Three consequences, in order of cost:

A defect below the cut has no symptom at all. The one that prompted this issue was found by reading the props schema, not by reading the build output.
The visible set is ordered by surface (pages, then views, then flows), so an entire surface can sit permanently below the cut on a repo whose page warnings alone exceed 50. On hotcrm today, every view and flow warning was invisible until a page warning was fixed.
Fixing warnings makes new ones appear, which reads as a regression introduced by the fix.

Neither the exit code nor the artifact is affected: both runs exit 0 and write dist/objectstack.json, as designed for advisory rules.

What would resolve it

Anything that keeps the printed set honest about its own truncation, in rough order of preference:

print all of them (the count is already known — the summary computes it);
keep a cap and print …and N more author-time warning(s) — rerun with --all naming the exact remainder;
keep a cap but make it per-surface, so no surface can be starved by another.

A silent cap is the one behaviour that should not survive, whatever the limit ends up being.

Found while fixing hotcrm#1248 (three record:related_list filters authored in shapes the component drops). Not fixed there: that repo consumes the CLI, it does not ship it.

Triage (Routine seat, hourly round): → pm:queue + domain:cli, type Bug.

Rationale: lands in packages/cli (build/validate warning printer). Dropping everything past 50 author-time warnings with no notice is silent truncation on an authoring feedback surface — the reader reads "50 warnings" as the whole set. Minimal honest fix (a "+N more" line or a flag) is dispatchable without a ruling; widening beyond that is not asked for.

Generated by Claude Code

Claim — `domain:cli` seat

Session session_019siH5jDmk5hrayvfyojUqR, branch claude/issue-11529-warning-truncation-notice. pm:queue → pm:dispatched + assignee in one stroke, read back.

Scope, from triage (5392389922): "Minimal honest fix (a +N more line or a flag) is dispatchable without a ruling; widening beyond that is not asked for."

The defect is not the cap — it is the silence. A reader who sees 50 warnings reads that as the whole set, on a surface whose entire job is authoring feedback. ⭐ That is the same shape as the family this lane closed twice this round (Runtime: dropping its row at zero, next: null overloading two causes): output that cannot distinguish "none" from "not shown".

⛔ Do not redesign the warning system. No new verbosity tiers, no paging, no config surface beyond what the minimal fix needs. If you find yourself wanting one, that is a finding to file, not scope to take.

Declared file surface: the build / validate warning printer in packages/cli and its test twin. ⚠️ In-flight #11485 is also in packages/cli (i18n-coverage.js + os i18n extract) — Ruling ① exempts same-package, but name your files precisely and ⛔ stop if you find yourself in theirs.

Clause-② — expected no (output gains an honesty line; no accept set, no public contract). Measure and say so.

Pins

Pin the discrimination, not the cap: over the limit, the output states how many were withheld; at or under it, no such line appears. A pin that only checks "50 printed" would pass on today's silent tree.

Anti-vacuity: shown failing without the fix. Prove mutations on disk by grepping both the injected and the removed text (an editor exit code proves nothing; a zero-hit sed exits 0), carry trap … EXIT INT TERM, predict the split before running, and report a wrong prediction rather than fitting it.

⚠️ A gate landed this morning: #11599's check:cli-test-child-env refuses a bulk process.env reference in any spawner under packages/cli/test/**. If your test spawns, use childEnv(). ⛔ Never add a baseline entry for new code — that baseline holds only leaks predating the gate.

Delivery

Changeset (user-visible output changes). Push, open a draft PR with Fixes #11529, run the repo's fast checks plus the packages/cli suite, derive the gate set with node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack, quote each gate's own verdict line. A gate that refuses is NOT MEASURED.

Generated by Claude Code

{
  "issue": 11529,
  "status": "done",
  "branch": "claude/issue-11529-warning-truncation-notice",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/11645",
  "premise_still_valid": true,
  "summary": "os build/compile printed a fixed 50 author-time advisories and stopped with no notice, so a truncated report was indistinguishable from a complete one. Extracted the advisory rendering out of compile.ts's inline loop into printAuthoringAdvisories() in packages/cli/src/utils/format.ts (with AuthoringAdvisory and AUTHORING_ADVISORY_PRINT_LIMIT = 50); over the cap it now prints '... and N more author-time warning(s) not shown (50 of 80) - re-run with --json for the full list', and at or under the cap the output is byte-for-byte unchanged. The pointer is --json, which already publishes the whole set (compile.ts:466) - no new flag, no verbosity tier, no paging, no config surface; the cap itself is untouched. build is an alias of compile, so compile.ts is the single printer for both. PREMISE CORRECTION: the card's secondary claim that 'objectstack validate prints the same truncated set' does NOT reproduce on origin/main 50fb191d - validate.ts collects every advisory into one warnings array (lines 268-301) and prints all of it uncapped at line 321; only its error lists are capped (140, 212) and both exit 1. The central premise (build silently truncates author-time warnings at 50) is confirmed at compile.ts:217 and is what this PR fixes; validate.ts is untouched and the changeset states the correction. Clause-2 measured 'no': packages/cli declares no exports map and src/index.ts re-exports command classes only (not utils/format.js), so the three new symbols are package-internal; no accept set moves, no schema/spec change, exit codes, the emitted artifact and all --json payloads unchanged - the only user-visible movement is one added line of stdout. Files touched: packages/cli/src/utils/format.ts, packages/cli/src/commands/compile.ts, packages/cli/test/build-warning-truncation-notice.test.ts, .changeset/build-warning-truncation-notice.md - none of them in-flight #11485's (i18n-coverage.js, os i18n extract).",
  "tests": "Union run at final commit 81dbb44a (git rev-parse --short HEAD), working tree clean; every heavy command through scripts/pm/os-verify-lock.sh and every result read from the wrapper's own VERDICT line, never a bare $?. NEW PIN packages/cli/test/build-warning-truncation-notice.test.ts - 7 tests, pinning the PAIR: over the limit the output states how many were withheld; at or under it no such line appears (a '50 printed' check alone passes on the silent tree). Green: 'Test Files 1 passed (1) / Tests 7 passed (7)'. ABLATION - prediction recorded BEFORE running: deleting the notice block reds exactly 3 (OVER the limit, ONE over the limit, the remainder is the EXACT count) and leaves 4 green (AT the limit, UNDER the limit, both controls). Observed exactly that: 'Test Files 1 failed (1) / Tests 3 failed | 4 passed (7)', the three predicted ones. Mutation proven ON DISK before the run, anchored on the exact text in both directions: injected marker=1 (want 1), removed notice line=0 (want 0). Script carried trap ... EXIT INT TERM; the trap fired from the wrong cwd and did NOT restore - caught by re-reading the file, restored explicitly (marker=0 notice=1, git status clean) and re-measured green 7/7. NO REBUILD LEG NEEDED and none claimed: the test resolves ../src/utils/format.js relatively into this package's own src/, so vitest transforms the mutated source directly - the ablation's dist condition (subject resolved through a dependency's exports) does not apply, and no mutated artifact could survive into a later run. PACKAGE CHECKS: pnpm --filter @objectstack/cli typecheck (script name echoed, 'tsc --noEmit', clean) and build - 'VERDICT command-exit 0'. FULL packages/cli suite in three shards to fit the container's ~10min foreground cap, all green: 57 files/783 tests, 56/565, 56/567 = 169 files, 1915 tests, 0 failures. NO NARROWING. GATE SET derived with 'node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack' (script read the change set from git itself: 4 paths vs merge base 50fb191dc; --repo assertion held against this checkout's origin), not a hand-built list. All green, each quoted by its own verdict line: check:nul-bytes 'OK (scanned 6504 text file(s) ... no raw ASCII control bytes)'; check:cli-test-child-env '28 spawner source(s) among 84 under packages/cli/test/**; no new bulk process.env copy reaches a spawned child (18 baselined, SHRINK-ONLY)' - the new test spawns nothing and NO baseline entry was added or edited; check:cross-package-test-inputs 'OK: 16 package(s) read outside themselves, all declared'; check:test-source-alias 'OK - 72 packages with tests scanned'; check:type-source-resolution 'OK - 77 packages'; check:published-files 'OK - 69 publishable package(s) of 78'; check:slot-lookup 'ratchet holds: 107 unswept site(s), none new'; check:query-options-erasure 'ratchet holds: 67 unswept non-test site(s), none new'; check:where-matcher '292 matcher(s) ... none new'; check:engine-double-contract 'OK - 397 pinned, 133 DEBT, 2 exempt'; check:plugin-teardown-shape '63 Plugin implementation(s) across 4584 source(s)'; check-ci-filter-parity 'OK: all 95 declared cross-package glob(s)'; check:changeset-gate-self-tests all three self-tests pass; check-adr-0087-registration 'no declared-breaking changeset (1 non-breaking seen)'; check-changeset-no-major 'no major bump'; check-empty-changeset 'No empty-frontmatter changeset (1 declaring changeset added)'; check:objectui-changeset 'objectui-range --self-test: all checks passed'; release-rehearsal-clone --self-test 'self-test passed'; docs-audit/check-affected-docs 'affected-docs self-test: 381 cases pass'; check:type-check-coverage exit 0, no ledger layer moved; check:i18n 'check-i18n-bundles: OK (9 package(s) - all bundles in sync)'. TWO GATES REFUSED FIRST AND ARE REPORTED AS RE-RUN, NOT AS PASSES: check:i18n-coverage returned 'COULD NOT MEASURE - 1 of 12 config(s) failed to lint' (@objectstack/connector-mcp had no build output in this worktree) and check:type-check-debt needs the workspace closure built. Built it - 'pnpm exec turbo run build --concurrency=2 --filter=./packages/* --filter=./packages/*/*', 'Tasks: 70 successful, 70 total', git status clean afterwards (no generated-artifact drift) - then re-ran both to real verdicts: check:i18n-coverage 'OK (12 config(s), 602 baselined untranslated string(s), none new)' and check:type-check-debt 'check-type-check-coverage --re-measure: OK - 32 ledger entr(ies) re-measured in 274.4s, 1897 raw tsc error(s) total, none above its recorded number' (its two informational 'can be lowered' surplus notes are for plugin-approvals and runtime, pre-existing and untouched by this diff). One note on the first full-build attempt: it was SIGTERM'd at the container's ~10min foreground cap (exit 143) and the two ELIFECYCLE lines in that log sit immediately before 'Terminated' - the kill cascade, not real build failures; the resumed run finished 70/70 green.",
  "open_questions": [],
  "out_of_scope_findings": [
    "filed as #11642: eight more silent 50-entry truncations in the same CLI (compile.ts 235/293/333/360, validate.ts 140/212, init.ts 893/908) - two are the same advisory-class defect, and compile.ts:293 even prints the true total in its header before showing 50; left alone because triage scoped this card to the author-time warning printer and said widening was not asked for",
    "filed as #11643: os build --json omits the undeclared-authoring-key warnings that os validate --json carries (compile.ts:291 guards them behind !flags.json, and the success payload at 466 sends ruleAdvisories only) - this is why the new notice points at --json for the authoring-rule advisories only; for that second list the pointer would be false, so it was left without a notice rather than given a broken remedy"
  ]
}

Generated by Claude Code

fix(cli): name the remainder at every truncating render in build/vali…

feat(cli): carry the computed advisory lists on every `os build --jso…

fix(cli): os build's text face renders every author-time advisory its…

Navigation Menu

Repository navigation

`objectstack build` / `validate` print at most 50 author-time warnings and drop the rest with no notice #11529

Measured

Why it matters

What would resolve it

Claim — `domain:cli` seat

Pins

Delivery

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

objectstack build / validate print at most 50 author-time warnings and drop the rest with no notice #11529

Description

Measured

Why it matters

What would resolve it

Activity

os-zhuang commented on Aug 24, 2026

os-zhuang commented on Aug 24, 2026

Claim — domain:cli seat

Pins

Delivery

claude commented on Aug 24, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

`objectstack build` / `validate` print at most 50 author-time warnings and drop the rest with no notice #11529

Claim — `domain:cli` seat