fix(types): SchemaRegistry's kanban entry stops describing a component it cannot name #10074

Workflow file for this run

.github/workflows/performance-budget.yml at 55a7470

	name: Bundle Analysis

	on:
	push:
	branches: [main, develop]
	paths:
	- 'packages/**'
	- 'apps/console/**'
	- 'pnpm-lock.yaml'
	# This file itself (objectui#6245). Without it a change to this gate ships
	# to `main` having never run once: the PR that edits it touches no
	# `packages/**` path, so `Bundle Analysis` never appears on it, and a
	# wiring bug surfaces on the NEXT `packages/**` PR — a required context
	# turning red on someone else's diff, reading to them as a bundle problem
	# of their own. Self-inclusion is this repo's convention for a
	# path-filtered gate: measured on `origin/main`, 5 of the 7 workflows
	# carrying a `paths:` filter list their own file, and this was one of the
	# two that did not. The cost is honest and accepted — every edit of this
	# file now pays a console build — and it is the same tax the other
	# self-included gates already pay.
	- '.github/workflows/performance-budget.yml'
	# ...and the gate's RUNTIME CLOSURE: the files this job actually executes.
	# Listing the YAML alone was still not enough — a PR touching only
	# `check-eager-closure-budget.mjs`, the file that computes the verdict and
	# the one this card is about, did not trigger this gate. The convention
	# measured on `origin/main` is the closure, not just the workflow:
	# `half-state-patrol.yml` lists its script AND `scripts/invoked-as.mjs`,
	# the dependency of that script.
	#
	# `invoked-as.mjs` is here for the same reason it is there, and the reason
	# is specific rather than tidy: `isEntrypoint` decides whether `main()`
	# runs at all, so a regression in it makes this checker exit 0 having
	# measured nothing — the exact silent-green failure the rest of this gate
	# exists to prevent.
	#
	# NOT listed: `scripts/__tests__/*.test.ts`. This job never runs vitest —
	# it runs two `node scripts/*.mjs` commands — so a test-only edit cannot
	# change this gate's verdict, and those tests already run on every PR in
	# the root vitest `unit` project. `published-dist-gate.yml` and
	# `spec-range-floors.yml` draw the same line.
	- 'scripts/check-eager-closure-budget.mjs'
	- 'scripts/render-budget-comment.mjs'
	- 'scripts/invoked-as.mjs'
	# The SDUI registration pins (objectui#6683) and their two dependencies.
	# Same rule as the three above — this job runs them, so an edit to any of
	# them must be able to turn this gate red on its OWN PR rather than on
	# somebody else's next `packages/**` change.
	- 'scripts/check-sdui-registration-pins.mjs'
	- 'scripts/check-side-effects-array.mjs'
	- 'scripts/component-registrations.mjs'
	- 'scripts/js-comment-mask.mjs'
	pull_request:
	branches: [main, develop]
	paths:
	- 'packages/**'
	- 'apps/console/**'
	- 'pnpm-lock.yaml'
	# This file itself (objectui#6245). Without it a change to this gate ships
	# to `main` having never run once: the PR that edits it touches no
	# `packages/**` path, so `Bundle Analysis` never appears on it, and a
	# wiring bug surfaces on the NEXT `packages/**` PR — a required context
	# turning red on someone else's diff, reading to them as a bundle problem
	# of their own. Self-inclusion is this repo's convention for a
	# path-filtered gate: measured on `origin/main`, 5 of the 7 workflows
	# carrying a `paths:` filter list their own file, and this was one of the
	# two that did not. The cost is honest and accepted — every edit of this
	# file now pays a console build — and it is the same tax the other
	# self-included gates already pay.
	- '.github/workflows/performance-budget.yml'
	# ...and the gate's RUNTIME CLOSURE: the files this job actually executes.
	# Listing the YAML alone was still not enough — a PR touching only
	# `check-eager-closure-budget.mjs`, the file that computes the verdict and
	# the one this card is about, did not trigger this gate. The convention
	# measured on `origin/main` is the closure, not just the workflow:
	# `half-state-patrol.yml` lists its script AND `scripts/invoked-as.mjs`,
	# the dependency of that script.
	#
	# `invoked-as.mjs` is here for the same reason it is there, and the reason
	# is specific rather than tidy: `isEntrypoint` decides whether `main()`
	# runs at all, so a regression in it makes this checker exit 0 having
	# measured nothing — the exact silent-green failure the rest of this gate
	# exists to prevent.
	#
	# NOT listed: `scripts/__tests__/*.test.ts`. This job never runs vitest —
	# it runs two `node scripts/*.mjs` commands — so a test-only edit cannot
	# change this gate's verdict, and those tests already run on every PR in
	# the root vitest `unit` project. `published-dist-gate.yml` and
	# `spec-range-floors.yml` draw the same line.
	- 'scripts/check-eager-closure-budget.mjs'
	- 'scripts/render-budget-comment.mjs'
	- 'scripts/invoked-as.mjs'
	# The SDUI registration pins (objectui#6683) and their two dependencies.
	# Same rule as the three above — this job runs them, so an edit to any of
	# them must be able to turn this gate red on its OWN PR rather than on
	# somebody else's next `packages/**` change.
	- 'scripts/check-sdui-registration-pins.mjs'
	- 'scripts/check-side-effects-array.mjs'
	- 'scripts/component-registrations.mjs'
	- 'scripts/js-comment-mask.mjs'

	concurrency:
	group: bundle-analysis-${{ github.event.pull_request.number \|\| github.ref }}
	cancel-in-progress: true

	permissions:
	contents: read
	pull-requests: write

	jobs:
	bundle-analysis:
	name: Bundle Analysis
	runs-on: ubuntu-latest
	steps:
	- name: Checkout code
	uses: actions/checkout@v7
	with:
	submodules: true
	# Depth 2, not 1: on a `pull_request` run HEAD is the MERGE REF, whose
	# first parent is the base commit this checkout was computed against.
	# The ceiling-freshness half (objectui#6245) needs that commit's copy
	# of the checker to tell a stale checkout from a PR that re-baselines
	# on purpose, and one extra commit is cheaper than a second fetch.
	fetch-depth: 2

	- name: Enable Corepack
	run: corepack enable

	- name: Verify pnpm version
	run: pnpm --version

	- name: Setup Node.js
	uses: actions/setup-node@v7
	with:
	node-version: '22.x'
	cache: 'pnpm'

	# ── Cache bookkeeping cannot void a recorded verdict (objectui#7048) ──
	# THE ORDERING, carried here from the type-check split in `ci.yml`
	# (objectui#6577, PR #7047) because it is what a future reader needs in
	# order to judge whether these steps may be touched: the verdict is
	# recorded by the checking steps; everything after them is bookkeeping,
	# and bookkeeping must never discard an answer the gate already produced.
	#
	# Why the SPLIT rather than a timeout on one step: combined
	# `actions/cache` declares `main: dist/restore/index.js` plus
	# `post: dist/save/index.js`, so its save is a step the RUNNER generates
	# at job end (`Post Turbo Cache`). No workflow syntax attaches
	# `timeout-minutes` or `continue-on-error` to a generated post step.
	# Measured once, on `ci.yml`'s type-check cache: a 1-second save took
	# 789s, ran that job into its ceiling, and the resulting `cancelled` —
	# which the merge queue cannot tell from `failure` — ejected an all-green
	# pull request (objectui#6577). `actions/cache`'s own `save-always`
	# deprecation text points at this same split.
	#
	# ⚠️ This job declares NO `timeout-minutes`, so its ceiling is GitHub's
	# default of 360 minutes, and `Bundle Analysis` is a required context
	# (objectui#6245). A stalled save would therefore hold a required check
	# open for up to six hours before reporting `cancelled`. ⛔ Adding a
	# ceiling is NOT the fix and is not attempted here — it would change what
	# this gate rejects. The bound below is on the bookkeeping step alone.
	#
	# The restore half is deliberately left UNBOUNDED: a restore stall fails
	# BEFORE any verdict exists — a gate that did not run, which is honest —
	# rather than a recorded verdict discarded.
	- name: Restore Turbo Cache
	id: turbo-cache
	uses: actions/cache/restore@v6
	with:
	path: .turbo/cache
	key: turbo-${{ runner.os }}-${{ github.sha }}
	restore-keys: \|
	turbo-${{ runner.os }}-

	- name: Install dependencies
	run: pnpm install --frozen-lockfile

	- name: Build packages
	id: build_packages
	run: pnpm turbo run build --filter='./packages/*'

	- name: Build Console
	run: pnpm --filter @object-ui/console build

	# objectui#6245: `Bundle Analysis` is a required context, and GitHub does
	# not re-run a PR's checks when the base branch moves — so a green verdict
	# can be computed against ceiling constants `main` has since replaced, and
	# the merge is then gated on a ceiling that no longer exists. Observed
	# live: run 32804357171 started at 03:13:27Z, 6m50s after `0409b766d`
	# lowered MAX_EAGER_CLOSURE_GZIP_BYTES from 4,086,000 to 3,345,000, and
	# published `BUDGET_CLOSURE_BUDGET_KB: 3990.2` — the retired ceiling — as a
	# success.
	#
	# This step hands the checker THREE readings of the constants: the two
	# below plus its own checkout. Two would not do — a re-baseline PR differs
	# from the base branch deliberately and has to stay landable — so the
	# verdict fires only when the base branch moved a ceiling AND this checkout
	# does not carry the move. See `evaluateCeilingFreshness`.
	#
	# `continue-on-error` is deliberate and is NOT a fail-open: when this step
	# cannot produce both files the checker finds the variables unset and
	# reports a freshness ERROR (exit 2), which fails the job. Letting the step
	# itself fail the job would kill it before the bundle is measured, and a
	# run that hides its numbers is how this gate loses its readers
	# (objectui#3152).
	#
	# The paths go out through `$GITHUB_ENV` rather than a step `env:` block:
	# `render-budget-comment.test.ts` walks every `NAME: ${{ steps.* }}` line
	# in this file and requires the COMMENT RENDERER to read it, which is the
	# right rule for the comment's inputs and the wrong one for the checker's.
	- name: Resolve the base-branch ceiling constants
	id: base_ceilings
	if: ${{ github.event_name == 'pull_request' }}
	continue-on-error: true
	env:
	PR_BASE_SHA: ${{ github.event.pull_request.base.sha }}
	BASE_REF: ${{ github.base_ref }}
	run: \|
	set -euo pipefail
	CHECKER='scripts/check-eager-closure-budget.mjs'
	OUT="${RUNNER_TEMP}/eager-closure-ceilings"
	mkdir -p "$OUT"

	# The base commit the merge ref was computed against. `fetch-depth: 2`
	# normally makes it a local object already; the fetch is the fallback
	# for a merge ref GitHub recomputed against a different base.
	if ! git cat-file -e "${PR_BASE_SHA}^{commit}" 2>/dev/null; then
	git fetch --no-tags --depth=1 origin "$PR_BASE_SHA"
	fi
	git cat-file -p "${PR_BASE_SHA}:${CHECKER}" > "$OUT/pr-base.partial"
	mv "$OUT/pr-base.partial" "$OUT/pr-base.mjs"

	# The base branch as it stands right now — what the merge would land on.
	git fetch --no-tags --depth=1 origin "+refs/heads/${BASE_REF}:refs/remotes/origin/${BASE_REF}"
	BASE_SHA=$(git rev-parse "refs/remotes/origin/${BASE_REF}")
	git cat-file -p "${BASE_SHA}:${CHECKER}" > "$OUT/base.partial"
	mv "$OUT/base.partial" "$OUT/base.mjs"

	# Written only after BOTH `mv`s: a half-populated pair would let the
	# checker compare against a file it thinks is the base branch.
	{
	echo "EAGER_CLOSURE_PR_BASE_SOURCE=$OUT/pr-base.mjs"
	echo "EAGER_CLOSURE_BASE_SOURCE=$OUT/base.mjs"
	echo "EAGER_CLOSURE_PR_BASE_SHA=$PR_BASE_SHA"
	echo "EAGER_CLOSURE_BASE_SHA=$BASE_SHA"
	echo "EAGER_CLOSURE_BASE_REF=$BASE_REF"
	} >> "$GITHUB_ENV"

	echo "Ceilings resolved: ${PR_BASE_SHA} (this checkout's base) vs ${BASE_SHA} (${BASE_REF} now)"

	- name: Check console performance budget
	id: budget
	run: \|
	# TWO measurements, one verdict.
	#
	# 1. The `index-*.js` entry chunk against a 350 KB line. This is the
	# original budget and it stays: a fat entry chunk is still worth
	# its own signal, and replacing a gauge is not licence to weaken
	# the one already here.
	# 2. The EAGER CLOSURE — every chunk the entry reaches through STATIC
	# imports, i.e. everything the browser must fetch and parse before
	# the app renders.
	#
	# Measurement 1 alone was the whole budget until objectui#5324. On
	# `77f846a8b` the entry chunk gzips to 25.9 KB against the 350 KB line
	# while the closure it pulls in is 3,881,609 bytes across 58 of 507
	# chunks — so the gate passed on 0.67% of the payload it claimed to
	# govern, and the 89 KiB regression of objectui#5266 landed in
	# `vendor-objectstack-*.js` where nothing here could see it.
	# `advancedChunks` routes vendor and workspace code into named chunks
	# on purpose, so MOST regressions land outside `index-*.js`.
	#
	# The closure ceiling is NOT set here. It lives in
	# `scripts/check-eager-closure-budget.mjs` next to the reasoning that
	# produced it, so a re-baseline is one edit against a documented
	# argument rather than a number in YAML: today's measured payload plus
	# half the 89 KiB regression the gate exists to catch, which keeps the
	# headroom NARROWER than that regression while leaving as much room for
	# the payload to shrink as to grow. Since objectui#5924 the checker
	# enforces that itself, against the report it just read rather than
	# against a frozen constant, for the aggregate ceiling and each
	# per-chunk one — a ceiling that has drifted out of range of the
	# regression it must catch is exit 2, a verdict about the gauge. It is
	# a truthful current-state ceiling, not a statement that 3.15 MB is fine.
	#
	# Both measurements run before either may fail the step: a run that
	# reports one number and hides the other teaches readers to distrust
	# the comment, which is how objectui#3152 nearly took this gate down.
	MAX_ENTRY_GZIP_KB=350

	DIST_DIR="apps/console/dist/assets"
	if [ ! -d "$DIST_DIR" ]; then
	echo "❌ Build output not found at $DIST_DIR"
	# Declare the outcome instead of leaving every output empty. An
	# empty `budget_status` used to render as a ❌ FAIL verdict
	# downstream (objectui#3152); the comment renderer now needs each
	# path to say what happened rather than infer it from silence.
	echo "budget_status=error" >> "$GITHUB_OUTPUT"
	echo "budget_message=Build output not found at $DIST_DIR" >> "$GITHUB_OUTPUT"
	exit 1
	fi

	# Find the main entry chunk (index-*.js)
	ENTRY_FILE=$(find "$DIST_DIR" -name 'index-.js' -not -name '.gz' -not -name '*.br' \| head -1)
	if [ -z "$ENTRY_FILE" ]; then
	echo "⚠️ Could not find main entry chunk, checking all JS files..."
	ENTRY_FILE=$(find "$DIST_DIR" -name '.js' -not -name '.gz' -not -name '*.br' \| sort \| head -1)
	fi

	if [ -z "$ENTRY_FILE" ]; then
	echo "❌ No JS files found in $DIST_DIR"
	echo "budget_status=error" >> "$GITHUB_OUTPUT"
	echo "budget_message=No JS files found in $DIST_DIR" >> "$GITHUB_OUTPUT"
	exit 1
	fi

	echo "📦 Main entry file: $(basename $ENTRY_FILE)"

	# Calculate gzip size
	GZIP_BYTES=$(gzip -c "$ENTRY_FILE" \| wc -c)
	GZIP_KB=$(awk "BEGIN {printf \"%.1f\", $GZIP_BYTES / 1024}")

	echo " Raw size: $(awk "BEGIN {printf \"%.1f\", $(wc -c < "$ENTRY_FILE") / 1024}") KB"
	echo " Gzip size: ${GZIP_KB} KB"
	echo " Budget: ${MAX_ENTRY_GZIP_KB} KB"

	echo "gzip_kb=$GZIP_KB" >> "$GITHUB_OUTPUT"
	echo "budget_kb=$MAX_ENTRY_GZIP_KB" >> "$GITHUB_OUTPUT"
	echo "entry_file=$(basename $ENTRY_FILE)" >> "$GITHUB_OUTPUT"

	ENTRY_OVER=$(awk "BEGIN {print ($GZIP_KB > $MAX_ENTRY_GZIP_KB) ? 1 : 0}")
	if [ "$ENTRY_OVER" -eq 1 ]; then
	echo "❌ ENTRY BUDGET EXCEEDED: Main entry is ${GZIP_KB} KB gzip (limit: ${MAX_ENTRY_GZIP_KB} KB)"
	else
	echo "✅ Entry budget OK: Main entry is ${GZIP_KB} KB gzip (limit: ${MAX_ENTRY_GZIP_KB} KB)"
	fi

	echo ""
	echo "📦 Eager closure (what a page load actually pays for):"
	# Writes closure_status / closure_gzip_kb / closure_budget_kb /
	# closure_chunks / closure_chunk_status / closure_headroom_status to
	# $GITHUB_OUTPUT itself. EVERY key it publishes is passed into the
	# comment step below and rendered there — a verdict published to
	# $GITHUB_OUTPUT that no consumer reads is a verdict the PR comment
	# cannot name, which is what objectui#6230 was (the per-chunk and
	# sensitivity halves both went unread). A fourth half added here must
	# be wired through too; `render-budget-comment.test.ts` fails if it is
	# not. The fourth half — ceiling freshness, objectui#6245 — arrived
	# through exactly that door: it publishes `closure_freshness_status`,
	# and is wired into the comment step below.
	#
	# Exit codes are distinct on
	# purpose: 1 = over budget (a verdict about the bundle), 2 = no
	# trustworthy measurement (a verdict about the gauge). Reporting one as
	# the other is how a broken gauge gets read as a size regression, and a
	# size regression as a broken gauge.
	set +e
	node scripts/check-eager-closure-budget.mjs
	CLOSURE_CODE=$?
	set -e

	if [ "$CLOSURE_CODE" -eq 2 ]; then
	echo "budget_status=error" >> "$GITHUB_OUTPUT"
	echo "budget_message=The entry chunk measured ${GZIP_KB} KB, but the eager-closure half of this gate returned no trustworthy VERDICT: the report could not be read, a ceiling has drifted out of range of the regression it must catch, or (objectui#6245) a ceiling was replaced on the base branch after this checkout was made. The step log says which. This is not a passing budget — and it is not a size regression either." >> "$GITHUB_OUTPUT"
	exit 1
	fi

	if [ "$ENTRY_OVER" -eq 1 ] \|\| [ "$CLOSURE_CODE" -ne 0 ]; then
	echo ""
	echo "❌ BUDGET EXCEEDED"
	echo "budget_status=fail" >> "$GITHUB_OUTPUT"
	exit 1
	fi

	echo ""
	echo "✅ Budget OK: entry chunk and eager closure are both within budget"
	echo "budget_status=pass" >> "$GITHUB_OUTPUT"

	# The ARTIFACT half of the `sideEffects` contract (objectui#6683). The
	# static gate in `ci.yml` proves the array agrees with the module bodies;
	# it cannot prove that a real bundler reading that array still EMITS the
	# registrations, and that is the question the hazard turns on —
	# `"sideEffects": false` is statically coherent and drops three live SDUI
	# widget registrations to 0 chunks on a green build (objectui#6535).
	#
	# The key set is derived from the array itself, so this step needs no
	# list of its own. It runs after the budget step and never instead of it:
	# a dropped registration and a size regression are different verdicts, and
	# a size win bought by deleting a feature must not read as a size win.
	#
	# Not `if: always()` — with no `dist/` the checker exits 2 and says so,
	# which is the honest verdict for a run that measured nothing, but there
	# is no reason to spend it on a build that already failed.
	- name: Pin the SDUI registrations the `sideEffects` array promises
	run: pnpm check:sdui-registration-pins

	- name: Generate package size report
	id: size-report
	# NOT `always()`. `always()` also fires on a cancelled run, where
	# `packages/*/dist` is only partly populated — the report was then
	# emitted from that partial tree and looked complete while silently
	# missing 7 packages (objectui#3152). The report is only meaningful
	# once every package has finished building, so gate it on that.
	if: ${{ !cancelled() && steps.build_packages.outcome == 'success' }}
	run: \|
	echo "## 📦 Bundle Size Report" > size-report.md
	echo "" >> size-report.md
	echo "\| Package \| Size \| Gzipped \|" >> size-report.md
	echo "\|---------\|------\|---------\|" >> size-report.md

	for pkg in packages/*/dist; do
	if [ -d "$pkg" ]; then
	pkg_name=$(basename $(dirname $pkg))

	# Calculate sizes for main bundle files
	for file in "$pkg"/*.js; do
	if [ -f "$file" ] && [ ! -f "${file}.map" ]; then
	# Use portable method to get file size
	size=$(wc -c < "$file")
	size_kb=$(awk "BEGIN {printf \"%.2f\", $size/1024}")

	# Estimate gzipped size
	gzip_size=$(gzip -c "$file" \| wc -c)
	gzip_kb=$(awk "BEGIN {printf \"%.2f\", $gzip_size/1024}")

	echo "\| $pkg_name ($(basename $file)) \| ${size_kb}KB \| ${gzip_kb}KB \|" >> size-report.md
	fi
	done
	fi
	done

	echo "" >> size-report.md
	echo "### Size Limits" >> size-report.md
	echo "- ✅ Core packages should be < 50KB gzipped" >> size-report.md
	echo "- ✅ Component packages should be < 100KB gzipped" >> size-report.md
	echo "- ⚠️ Plugin packages should be < 150KB gzipped" >> size-report.md

	# Rendering lives in `scripts/render-budget-comment.mjs` so it can be unit
	# tested (`scripts/__tests__/render-budget-comment.test.ts`) — the bug this
	# replaces was purely a rendering bug, and logic inlined in YAML is
	# untestable. Step outputs are passed through `env` rather than
	# interpolated into a JS string literal, so an absent output stays an
	# empty variable instead of vanishing into source text.
	- name: Render performance budget comment
	id: render_comment
	# `!cancelled()`, NOT `always()`: a cancelled run measured nothing, and
	# the superseding run posts the real verdict moments later. Commenting
	# on cancellation is what produced a ❌ FAIL on every PR that got a
	# second push (objectui#3152).
	if: ${{ github.event_name == 'pull_request' && !cancelled() }}
	env:
	BUDGET_STATUS: ${{ steps.budget.outputs.budget_status }}
	BUDGET_MESSAGE: ${{ steps.budget.outputs.budget_message }}
	BUDGET_GZIP_KB: ${{ steps.budget.outputs.gzip_kb }}
	BUDGET_LIMIT_KB: ${{ steps.budget.outputs.budget_kb }}
	BUDGET_ENTRY_FILE: ${{ steps.budget.outputs.entry_file }}
	BUDGET_CLOSURE_STATUS: ${{ steps.budget.outputs.closure_status }}
	BUDGET_CLOSURE_GZIP_KB: ${{ steps.budget.outputs.closure_gzip_kb }}
	BUDGET_CLOSURE_BUDGET_KB: ${{ steps.budget.outputs.closure_budget_kb }}
	BUDGET_CLOSURE_CHUNKS: ${{ steps.budget.outputs.closure_chunks }}
	# The per-chunk (objectui#5490) and sensitivity (objectui#5924) halves.
	# The step's exit code folds all three halves into one `budget_status`,
	# so without these the comment can say a budget objected but not which
	# half did — the reader has to open the job log to find out.
	BUDGET_CLOSURE_CHUNK_STATUS: ${{ steps.budget.outputs.closure_chunk_status }}
	BUDGET_CLOSURE_HEADROOM_STATUS: ${{ steps.budget.outputs.closure_headroom_status }}
	# The freshness half (objectui#6245). Empty on a run it does not apply
	# to; the renderer filters an empty half out rather than rendering it.
	BUDGET_CLOSURE_FRESHNESS_STATUS: ${{ steps.budget.outputs.closure_freshness_status }}
	BUDGET_STEP_OUTCOME: ${{ steps.budget.outcome }}
	BUILD_PACKAGES_OUTCOME: ${{ steps.build_packages.outcome }}
	run: node scripts/render-budget-comment.mjs > budget-comment.md

	- name: Comment PR with results
	if: ${{ github.event_name == 'pull_request' && !cancelled() }}
	continue-on-error: true
	uses: actions/github-script@v9
	with:
	# GitHub's API intermittently rejects valid run tokens with 401
	# (seen on PR #1627), so 401/403 must stay retryable here.
	retries: 3
	retry-exempt-status-codes: 400,404,422
	script: \|
	const fs = require('fs');

	if (!fs.existsSync('budget-comment.md')) {
	core.warning('budget-comment.md was not rendered; skipping PR comment.');
	return;
	}
	const body = fs.readFileSync('budget-comment.md', 'utf8');

	try {
	await github.rest.issues.createComment({
	issue_number: context.issue.number,
	owner: context.repo.owner,
	repo: context.repo.repo,
	body
	});
	} catch (error) {
	core.warning(`Could not comment on PR: ${error.message}`);
	await core.summary.addRaw(body).write();
	}

	# The bookkeeping half of the split documented at the restore step above,
	# placed HERE — last in the job, after the budget verdict AND after the
	# comment that reports it — because that is exactly where the post phase
	# it replaces already ran. Nothing about which commits this gate accepts
	# or rejects moves with it.
	#
	# `timeout-minutes: 5` is DERIVED FOR THIS SITE, and landing on the same
	# number PR #7047 chose is a result rather than a copy — objectui#7048
	# fences inheriting that 5, so here is this site's own arithmetic. This
	# job's own save of this cache measured 1s on 2026-09-02 (`Post Turbo
	# Cache`, job 100096569559); the whole job was 4m07s. 5 minutes is 300x
	# the measured save — the same ratio #7047 sized its bound at, arrived at
	# from this job's own number.
	#
	# `continue-on-error: true` is the other half, and without it the bound
	# would only trade a `cancelled` gate for a red one. A cache that failed
	# to upload costs the next run some time; it says nothing whatsoever
	# about the bundle sizes under test, so it must not speak for them.
	#
	# Behaviour preserved, spelled out so the equivalence is checkable:
	# - `cache-hit != 'true'` reproduces the combined action's own "exact
	# hit on the primary key ⇒ do not save" skip. A `restore-keys` prefix
	# match leaves `cache-hit` false and the save still runs, exactly as
	# the combined action behaved.
	# - the condition names no status function, so the implicit `success()`
	# still applies — matching the combined action's `post-if: success()`.
	# - same `path` and same `key` as the restore step above.
	- name: Save Turbo Cache
	if: steps.turbo-cache.outputs.cache-hit != 'true'
	uses: actions/cache/save@v6
	timeout-minutes: 5
	continue-on-error: true
	with:
	path: .turbo/cache
	key: turbo-${{ runner.os }}-${{ github.sha }}

Navigation Menu

Uh oh!

fix(types): SchemaRegistry's kanban entry stops describing a component it cannot name #10074

Workflow file

fix(types): SchemaRegistry's kanban entry stops describing a component it cannot name #10074

Uh oh!

Workflow file for this run