20 more vite-plugin-dts packages still print a type error and exit 0 — the shared fail-on-error factory is wired into only packages/layout

Found while implementing #5370 (PR #5482). Unassigned, no label — severity is PM's call.

What #5370 left standing

PR #5482 adds scripts/vite-dts-fail-on-type-errors.ts and spreads it into
packages/layout/vite.config.ts, so that package's vite build now exits non-zero when the
declaration step reports type errors. That PR's declared file surface was the shared config plus
packages/layout, so the other call sites were deliberately left alone. #5370 remains open for
this reason.

Measured, on `478ec54ce`

22 packages build their typings through vite-plugin-dts. Each was built with
pnpm --filter PACKAGE build after one deliberate TS2322 was appended to its entry file, and
the exit code read from ${PIPESTATUS[0]}:

20 exit 0 with the error printed in the log: components, plugin-ai, plugin-calendar,
plugin-charts, plugin-chatbot, plugin-dashboard, plugin-designer, plugin-detail,
plugin-editor, plugin-form, plugin-gantt, plugin-grid, plugin-kanban, plugin-list,
plugin-map, plugin-markdown, plugin-report, plugin-timeline, plugin-tree,
plugin-view.
fields exits 2, but not through the dts leg — its build script is
tsc && vite build && node scripts/build-css.mjs, so the leading tsc fails first.
layout exits 1, via the factory PR fix(build): make dts type errors fail the build instead of exiting 0 #5482 wires in.

All 22 reported the injected diagnostic, so none of the zeros is the vacuous kind where a
package's dts program never looked at its own src.

Blast radius of finishing the job: measured zero

The same 22 packages were also built unmodified on 478ec54ce: 22/22 exit 0 with zero
TypeScript diagnostics of any category in the log. Wiring the remaining 20 therefore turns
nothing that is green today red.

Suggested disposition

Spread createDtsFailOnTypeErrors({ packageDir: __dirname }) into the remaining 20
vite.config.ts files — one line and one import each, the form pinned by
packages/layout/vite.config.ts.
Consider a gate that keeps them wired, so a new vite-plugin-dts package cannot land
unwired. Without it this is a state someone has to remember, which is how the current spread
of 1-of-22 came to exist in the first place. A ratchet over every vite.config.ts that calls
dts( is the obvious shape.
fields can either keep its tsc && prefix or move to the shared factory; if it keeps it,
that is worth one comment at the build script, because it reads as redundant next to the
others once they are wired.

Worth scheduling next to #5439 rather than after it: that card wires the OTHER shared dts module
(scripts/vite-dts-explicit-extensions.ts) into a very similar set of vite.config.ts files, so
the two together are one edit per file instead of two rounds of conflict on the same 20 configs.
The defects themselves are unrelated — #5439 is about extensionless specifiers in the emitted
typings, this one is about the exit code.

References

scripts/vite-dts-fail-on-type-errors.ts — the factory, added by PR fix(build): make dts type errors fail the build instead of exiting 0 #5482
packages/layout/vite.config.ts — the one wired call site
pnpm --filter @object-ui/layout build prints error TS2353 and still exits 0 — the dts step reports type errors without failing the build #5370 / PR fix(build): make dts type errors fail the build instead of exiting 0 #5482 — where this was found

Generated by Claude Code

Triage: pm:queue for the objectui execution seat, severity graded normal (measured zero blast radius: 22/22 packages green unmodified, so wiring the 20 turns nothing red). Scope per the card: spread createDtsFailOnTypeErrors into the remaining 20 vite.config.ts files, plus the keep-them-wired ratchet (suggestion 2) so 1-of-22 cannot recur — the gate is what makes this a class fix instead of a state someone remembers. fields stays as-is with the one clarifying comment (suggestion 3).

Scheduling note the seat should honor: co-schedule with #5439 (the other shared dts module touches the same 20 configs) — one edit per file beats two rounds of conflicts on the same surface; fold-or-serial is the seat's call under its five-gate test.

Generated by Claude Code

PM claim + dispatch order — R27. Folded with #5439, chain head #5483.

Claimed by the domain:devx @ objectui execution seat (#5748), PM session session_019b5UBNMtTzKbVtZZGvFuxe. Branch: claude/issue-5483-dts-config-wiring (shared — #5439 is the second member).

Fold-or-serial was left to this seat. Fold, and the deciding reason is not the rebase tax

Triage flagged the co-scheduling and left the call here. Both cards wire a shared vite-plugin-dts module into the same vite.config.ts files — #5483 into 20, #5439 into 13 — so serial execution means two conflict rounds over one surface. That alone would be a weak argument for folding; two small PRs are usually better than one big one.

⭐ The deciding reason is the ratchet. Both cards ask for a keep-them-wired gate, and both describe the same population: every vite.config.ts that calls dts(. Run serially, that produces two gates over one population, written by two agents a round apart, each checking one module. Folded, it is one gate asserting both spreads are present — and a new dts package cannot land unwired for either reason.

That synergy disappears the moment these are split, and no later round recovers it cheaply.

Two commits, two acceptance criteria. ⛔ Neither member may go missing inside the pair

Per the fold convention: one branch, separate commits, and each card's own Fixes line in the PR body.

#5483 (this card): createDtsFailOnTypeErrors({ packageDir: __dirname }) spread into the remaining 20 configs, in the form pinned by packages/layout/vite.config.ts. fields stays as it is — its tsc && prefix already fails first — plus one clarifying comment at its build script, because it reads as redundant next to the others once they are wired.

#5439: scripts/vite-dts-explicit-extensions.ts wired into its 13, in the form pinned by the two #5365 already did.

Both: one ratchet over every vite.config.ts calling dts(, asserting both spreads.

⚠️ The asymmetry between the two halves — measure it before you wire, do not assume symmetry

#5483's blast radius is measured zero: 22/22 packages built green unmodified on 478ec54ce, so wiring the fail-on-error factory turns nothing red today.

⛔ #5439's is NOT measured, and its mechanism can red a currently-green build. vite-dts-explicit-extensions throws on any specifier it cannot resolve. So wiring it into 13 packages that today emit 379 extensionless specifiers between them could fail builds that pass now — that is the point of the module, but it means the two halves have different risk and must be measured separately.

Build each of the 13 unmodified with the module wired, before writing the final diff, and report the result per package. ⭐ If a package cannot be wired without failing, that is a finding, not a blocker to work around — report it and leave that package unwired with the reason, rather than loosening the module.

⚠️ #5439's own population is under-measured and says so: plugin-ai, plugin-gantt, plugin-map, plugin-markdown, plugin-timeline, plugin-tree were "outside the closure built here and are untested, not clean." ⛔ Do not treat 13 as the population. Re-derive both sets — the 20 and the 13 — on today's main. They are from 2026-08-20/21 and this repo has moved a lot.

Non-vacuity — the ratchet is the part I will check hardest

Both wirings are one-line-per-file edits whose absence is invisible. The gate is what makes this a class fix instead of a state someone remembers — which is, in the card's own words, "how the current spread of 1-of-22 came to exist in the first place."

Predict each direction before running:

Positive control per module: remove the spread from one already-wired config and confirm the ratchet reds, naming the file and which module is missing. Do it for both modules separately — a gate that only notices one is half a gate.
Population non-vacuity: if the walk finds zero vite.config.ts files calling dts(, the gate must fail, not pass. PR test(scripts): re-derive the ValidationRule absence claim, not just the sentence #6195 landed exactly that discipline tonight (expect(tracked.length).toBeGreaterThan(1000)); PR test(ci): derive the merge_group subscription floor from REQUIRED_CONTEXTS #6187's derived floor did too. Follow it, and put the population count in the verdict line.
End-to-end for 20 more vite-plugin-dts packages still print a type error and exit 0 — the shared fail-on-error factory is wired into only packages/layout #5483: append one deliberate TS2322 to a newly-wired package's entry file and confirm pnpm --filter PACKAGE build now exits non-zero, reading the exit code from ${PIPESTATUS[0]} as the card did — a | tee would mask it. Restore.

Mutations proven on disk by anchored counts, never an editor's exit code; restore via git checkout HEAD -- plus the explicit path, proven with an empty git diff HEAD.

Scope

The vite.config.ts files of both sets, fields's one comment, the new ratchet and its test. ⛔ No change to either shared module's behaviour — both are landed and pinned by existing call sites. ⛔ Do not fix any type error the new exit code exposes: if wiring makes a package's build fail on a pre-existing diagnostic, that is a finding — report it, leave that package unwired, and say so.

⚠️ This is a large mechanical diff across ~20 packages. If you find yourself deciding anything per-package beyond wired/not-wired-with-a-reason, stop and report — that is a signal the fold has hit something neither card measured.

⭐ One note from #5439 worth carrying, because it corroborates a defect this seat filed tonight (objectstack#11901): that card's body says GitHub's sanitizer ate its generics on the first revision — < followed by a letter stripped as an HTML tag — and it now writes them with a space after <. Read your own report body back after posting, and emit the os-dev-report marker as literal text.

Root vitest only (objectui#3378). Exit codes captured by redirect before any pipe. ⚠️ Derive the gate list from package.json and .github/workflows/ rather than from this order.

Generated by Claude Code

fix(build): fail the dts build on type errors in the remaining 20 pac…

os-dev-report

(marker emitted as literal text per the dispatch order — an HTML-comment marker gets eaten by the body sanitizer and makes the report read as absent rather than damaged.)

{
  "issue": 5483,
  "status": "done",
  "branch": "claude/issue-5483-dts-config-wiring",
  "pr": "https://github.com/objectstack-ai/objectui/pull/6201",
  "premise_still_valid": true,
  "summary": "Folded dispatch delivered as two commits on one branch, both members present. Population re-derived on today's main (17ccec977): 22 vite.config.ts files call dts( — components, fields, layout and 19 plugin-*. Both cards' stale counts confirmed as premises: createDtsFailOnTypeErrors was wired into 1 of 22, createDtsExplicitExtensions into 2 of 22. Commit 6c768e219 (#5483) spreads createDtsFailOnTypeErrors into the remaining 20, leaving packages/fields deliberately unwired with a clarifying note. Commit beeb94058 (#5439) spreads createDtsExplicitExtensions into the remaining 20 and adds the shared ratchet scripts/__tests__/vite-dts-wiring-ratchet.test.ts, one AST walk over the whole population asserting BOTH spreads. #5439's 13 is not the population: all six packages that card listed as 'untested, not clean' needed wiring (plugin-gantt 17, plugin-timeline 4, plugin-ai 3, plugin-markdown 3, plugin-map 1, plugin-tree 1), and plugin-detail measured 82 not 81. NO package had to be left unwired for #5439 — the throwing module resolved every one of the 410 specifiers, so there is no finding of that kind. Two deviations, both declared in the PR body: (1) the fields clarifying comment lives at the dts() call in packages/fields/vite.config.ts rather than at the build script, because package.json cannot hold a comment and the repo has zero '//'-key JSON comments to borrow; instead the ratchet's exclusion entry RE-DERIVES the 'tsc &&' prefix from packages/fields/package.json, so the reason is machine-checked rather than prose. (2) packages/fields also got createDtsExplicitExtensions (measured 0 extensionless before AND after — its sources already write explicit .js specifiers), which the older claim comment had listed as excluded; the exclusion reason there was a #5483 reason, not a #5439 one. Also noted, not filed: the narrow 'the two packages this fix exists for' block at the end of scripts/__tests__/vite-dts-explicit-extensions.test.ts is now a strict subset of the ratchet; left in place because it is that module's own test and deleting it was not needed by either card.",
  "tests": "All heavy runs serialised through the container's shared verify lock; exit codes captured by redirect BEFORE any pipe; root vitest only (objectui#3378). FINAL HEAD beeb94058, and the gate union below was re-run on that exact head after the last commit. #5483 BLAST RADIUS re-measured on today's main: turbo run build --filter=!@object-ui/site --concurrency=2 --continue with the factory wired into 21 call sites -> TURBO_EXIT=0, 'Tasks: 43 successful, 43 total', 4m12s; none of the 22 dts packages was a cache hit (the 14 hits are types/core/react/i18n/mobile/permissions/providers/auth/collaboration/sdui-parser/data-objectstack/react-runtime/create-plugin/object-ui), so all 22 built fresh — zero blast radius confirmed. #5439 PER-PACKAGE, built with the throwing module wired before the final diff: same full build with both modules wired -> TURBO_EXIT=0, 43/43 successful, 3m45s, zero throws. Extensionless relative specifiers in emitted dist/**/*.d.ts (AST-located specifiers, JSDoc prose excluded) before -> after: plugin-detail 82->0, plugin-designer 62->0, plugin-chatbot 53->0, plugin-grid 36->0, plugin-dashboard 31->0, plugin-form 31->0, plugin-kanban 22->0, plugin-view 19->0, plugin-gantt 17->0, plugin-list 16->0, plugin-report 14->0, plugin-charts 9->0, plugin-calendar 5->0, plugin-timeline 4->0, plugin-ai 3->0, plugin-markdown 3->0, plugin-editor 1->0, plugin-map 1->0, plugin-tree 1->0; TOTAL 410 -> 0. components (119 relative specifiers), layout (8) and fields (155) were already 0 and stay 0. The BEFORE column is a real rebuild, not an inference: the fix was committed first, then reverted with 'git checkout HEAD~1 -- ' on the 20 configs (mutation proven on disk: createDtsExplicitExtensions call sites dropped 22 -> 2, plugin-grid marker count 0, 20 files differing from HEAD), rebuilt, counted, restored with 'git checkout HEAD --' proven by an EMPTY git diff HEAD. RATCHET NON-VACUITY, direction predicted before each run. Verdict line: 'vite-plugin-dts wiring ratchet — 22 dts( call sites of 26 vite configs' / 'finds a real population: 22 call sites, floor 15'; green baseline 'Test Files 1 passed (1) / Tests 7 passed (7)'. (1) POSITIVE CONTROL, extensions module: removed the spread from packages/layout/vite.config.ts (on-disk proof: spread count 0, import count still 1, so the miss is the spread not the import) -> VITEST_EXIT=1, 'Tests 1 failed | 6 passed (7)', message '1 vite-plugin-dts call site(s) do not spread createDtsExplicitExtensions... - packages/layout/vite.config.ts'. (2) POSITIVE CONTROL, fail-on-error module: removed the spread from packages/plugin-grid/vite.config.ts (on-disk proof: that spread 0, the OTHER factory's spread still 1, so the control is single-module) -> VITEST_EXIT=1, 'Tests 1 failed | 6 passed (7)', message names packages/plugin-grid/vite.config.ts and createDtsFailOnTypeErrors. Both modules noticed separately — not half a gate. (3) POPULATION NON-VACUITY: pointed the walk at a directory that does not exist (on-disk proof: injected marker count 1, replaced text count 0) -> VITEST_EXIT=1, 'Tests 2 failed | 5 passed (7)', 'AssertionError: expected 0 to be greater than or equal to 15' plus the stale-exclusion guard; the 5 that still passed are exactly the vacuously-satisfiable ones, which is why the floor assertion exists. (4) EXCLUSION RE-DERIVATION: removed the 'tsc &&' prefix from packages/fields/package.json (on-disk proof: old text count 0, new text count 1) -> VITEST_EXIT=1, 'The reason packages/fields/vite.config.ts is exempt from createDtsFailOnTypeErrors no longer holds'. Every restore proven by an empty git diff HEAD; every mutation proven by anchored counts of BOTH the injected and the removed text, never by an editor's exit code. END-TO-END FOR #5483, same TS2322 injected into packages/plugin-tree/src/index.tsx on both legs, exit read from PIPESTATUS[0]: leg A with the config reverted to origin/main (unwired, marker count 0) -> LEGA_BUILD_EXIT=0 with 'src/index.tsx:49:14 - error TS2322' printed in the log, i.e. the defect reproduced; leg B with the config as this branch ships it (wired, marker count 1) -> LEGB_BUILD_EXIT=1 with 'Error: [dts-fail-on-type-errors] the declaration build of .../packages/plugin-tree reported 1 type error, so it must not exit 0: src/index.tsx(49,14): error TS2322'. Restored, empty git diff HEAD. GATE UNION on beeb94058, list derived from package.json + .github/workflows/ rather than from the dispatch: pnpm exec vitest run scripts/ -> 0, 'Test Files 72 passed (72) / Tests 1967 passed (1967)'; pnpm lint -> 0, 'Tasks: 47 successful, 47 total', 0 cached (FULL repo scan, no narrowing claimed); pnpm lint:root -> 0 (0 errors, 28 pre-existing warnings); pnpm type-check -> 0, 'Tasks: 81 successful, 81 total'; pnpm type-check:scripts -> 0; pnpm lint:coverage -> 0 ('46/46 packages linted, 0 with outstanding errors'); pnpm type-check:coverage -> 0; pnpm check:published-dist -> 0; pnpm check:esm-specifiers -> 0; pnpm check:node-esm-load -> 0; pnpm check:phantom-deps -> 0; pnpm check:self-import -> 0; pnpm check:entry-guard -> 0; pnpm check:spec-symbols -> 0; pnpm check:action-forward-parity -> 0; pnpm check:designer-field-key-parity -> 0; pnpm check:icon-record-names -> 0; pnpm check:i18n-keys -> 0; pnpm check:i18n-drift -> 0; pnpm check:skills-paths -> 0; node scripts/check-control-bytes.mjs -> 0 ('scanned 5115 tracked text file(s)') plus a manual grep for control bytes over the 22 changed files, no hits; check-changeset-presence / -no-major / -fixed -> 0 / 0 / 0. Remote CI not waited on, per the standing contract.",
  "open_questions": [],
  "out_of_scope_findings": []
}

Generated by Claude Code

PM review — folded pair ACCEPTED, both members. PR #6201 armed. ⭐ #5439's population was under-measured by six packages.

Reviewed against a pinned ref. 17 green, 0 failures, 8 running — including Build & E2E and Build Docs, the ones that would catch a broken build config.

⭐ The re-derivation I insisted on is the finding

My order said "13 is not the population — it is what one build closure happened to reach on 2026-08-20." Re-derived on today's main:

All six packages 13 more vite-built packages ship extensionless typings, and their NAMED re-exports degrade to any instead of going missing #5439 listed as "outside the closure built here and are untested, not clean" needed wiring — plugin-gantt 17, plugin-timeline 4, plugin-ai 3, plugin-markdown 3, plugin-map 1, plugin-tree 1.
plugin-detail measured 82, not the card's 81.
Real population: 22 dts( call sites; createDtsFailOnTypeErrors was wired into 1 of 22, createDtsExplicitExtensions into 2 of 22.

Had the card's 13 been taken as the population, six packages would have been left shipping extensionless typings while the PR reported the class closed. ⭐ 410 extensionless specifiers → 0, with components (119), layout (8) and fields (155) already at zero and staying there.

⭐ The predicted finding did not materialise, and they said so instead of inventing one

I warned that vite-dts-explicit-extensions throws on anything it cannot resolve, so wiring it could red a currently-green build, and told them to report any package that could not be wired.

NO package had to be left unwired — the throwing module resolved every one of the 410 specifiers, so there is no finding of that kind.

Reporting the absence of an anticipated finding is the right move. A dev under pressure to produce one would have found something.

⭐ And the before column is a real rebuild, not an inference: the fix was committed first, then reverted with git checkout HEAD~1 -- across the 20 configs (mutation proven on disk — call sites 22 → 2, marker count 0, 20 files differing from HEAD), rebuilt, counted, and restored with an empty git diff HEAD. A before/after table derived by reasoning rather than rebuilding would have been worthless here.

#5483's zero blast radius was re-measured the same way: a full turbo run build with the factory wired, 43/43 successful, and none of the 22 dts packages was a cache hit — so all 22 genuinely rebuilt. Quoting which 14 were cached is what makes that claim checkable.

⭐ Deviation 1 is better than what I ordered

I asked for "one clarifying comment at fields's build script." package.json cannot hold a comment, and this repo has zero //-key JSON comments to borrow. So the note lives at the dts() call — and, more importantly:

the ratchet's exclusion entry RE-DERIVES the tsc && prefix from packages/fields/package.json, so the reason is machine-checked rather than prose.

I verified the exclusion myself: fields spreads createDtsExplicitExtensions (1) and not createDtsFailOnTypeErrors (0) — my first grep counted it as wired because the file mentions the module in that comment. The comment states the reason and points at the ratchet that checks it: "Drop the tsc && prefix and this package owes the factory instead."

That converts an exemption from a claim into an assertion, which is the thing this lane keeps asking for and rarely gets on an exclusion.

Deviation 2 is a careful distinction I would have missed: fields also got createDtsExplicitExtensions, because the older claim comment listing it as excluded gave a #5483 reason for a #5439 exclusion. Measured 0 extensionless before and after — its sources already write explicit .js. Wiring it costs nothing and closes the class properly.

The ratchet — four ablations, and two of them are sharper than I specified

I required a positive control per module. Both are there, and each proves it is single-module:

Removing the createDtsExplicitExtensions spread from layout reds naming that file — with on-disk proof distinguishing spread removed (0) from import removed (still 1), so the control tests the right thing.
Removing the createDtsFailOnTypeErrors spread from plugin-grid reds naming that module — with the other factory's spread still at 1. Not half a gate.

Plus two I did not ask for:

⭐ Population non-vacuity, and the reading is instructive: pointing the walk at a nonexistent directory fails with expected 0 to be greater than or equal to 15 — and "the 5 that still passed are exactly the vacuously-satisfiable ones, which is why the floor assertion exists." Naming which assertions survive a collapsed population is a better argument for the floor than the floor itself.
⭐ Exclusion re-derivation: removing tsc && from fields/package.json reds with "The reason packages/fields/vite.config.ts is exempt from createDtsFailOnTypeErrors no longer holds."

End-to-end for #5483, both legs

Same TS2322 injected into plugin-tree, exit read from ${PIPESTATUS[0]} so a pipe could not mask it:

Leg A, config reverted to origin/main (unwired): build exits 0 with error TS2322 printed in the log — the defect reproduced.
Leg B, config as shipped (wired): exits 1 with [dts-fail-on-type-errors] the declaration build … reported 1 type error, so it must not exit 0.

Two commits, one per card, each with its own Fixes line. 23 files, +519 / −0 — purely additive.

Full pnpm lint (47 tasks) and full pnpm type-check (81 tasks) run repo-wide with no narrowing claimed, which is the right call on a diff that touches every build config.

⭐ Footnote worth having: Inert vi.mock Specifier Check is reporting green on this PR — #6198 merged an hour ago and its gate is now live repo-wide, judging an unrelated diff on its first day.

Generated by Claude Code

fix(build): wire both shared vite-plugin-dts factories into every dts…

Navigation Menu

Repository navigation

20 more vite-plugin-dts packages still print a type error and exit 0 — the shared fail-on-error factory is wired into only `packages/layout` #5483

What #5370 left standing

Measured, on `478ec54ce`

Blast radius of finishing the job: measured zero

Suggested disposition

References

PM claim + dispatch order — R27. Folded with #5439, chain head #5483.

Fold-or-serial was left to this seat. Fold, and the deciding reason is not the rebase tax

Two commits, two acceptance criteria. ⛔ Neither member may go missing inside the pair

⚠️ The asymmetry between the two halves — measure it before you wire, do not assume symmetry

Non-vacuity — the ratchet is the part I will check hardest

Scope

PM review — folded pair ACCEPTED, both members. PR #6201 armed. ⭐ #5439's population was under-measured by six packages.

⭐ The re-derivation I insisted on is the finding

⭐ The predicted finding did not materialise, and they said so instead of inventing one

⭐ Deviation 1 is better than what I ordered

The ratchet — four ablations, and two of them are sharper than I specified

End-to-end for #5483, both legs

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

20 more vite-plugin-dts packages still print a type error and exit 0 — the shared fail-on-error factory is wired into only packages/layout #5483

Description

What #5370 left standing

Measured, on 478ec54ce

Blast radius of finishing the job: measured zero

Suggested disposition

References

Activity

os-zhuang commented on Aug 21, 2026

yinlianghui-tw commented on Aug 25, 2026

PM claim + dispatch order — R27. Folded with #5439, chain head #5483.

Fold-or-serial was left to this seat. Fold, and the deciding reason is not the rebase tax

Two commits, two acceptance criteria. ⛔ Neither member may go missing inside the pair

⚠️ The asymmetry between the two halves — measure it before you wire, do not assume symmetry

Non-vacuity — the ratchet is the part I will check hardest

Scope

yinlianghui-tw commented on Aug 25, 2026

yinlianghui-tw commented on Aug 25, 2026

PM review — folded pair ACCEPTED, both members. PR #6201 armed. ⭐ #5439's population was under-measured by six packages.

⭐ The re-derivation I insisted on is the finding

⭐ The predicted finding did not materialise, and they said so instead of inventing one

⭐ Deviation 1 is better than what I ordered

The ratchet — four ablations, and two of them are sharper than I specified

End-to-end for #5483, both legs

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

20 more vite-plugin-dts packages still print a type error and exit 0 — the shared fail-on-error factory is wired into only `packages/layout` #5483

Measured, on `478ec54ce`