fix: test: budget of test:changed follows the selection scope

AGENTS.md

@@ -54,7 +54,7 @@ Unit tests run on `bun test` (repo root, `bunfig.toml`); component and visual te
 
 | Command | What it runs | Budget | Measured |
 |---|---|---|---|
-| `bun run test` = `test:changed` | what the diff affects: `bun test --changed` (unit); the component tests whose imports reach a changed file (`scripts/test-graph.ts`: a static import graph with the barrel flattened, because Vitest's own `--changed` selects every file in browser mode); the screenshots of those components; the docs routes mapped by `scripts/route-impact.ts`. Prints what it selected and why. Every component test runs when a shared file changed (styles, tokens, theme, stores, primitives, a test setup file). The site part needs a fresh `site/dist` (a build is ~75 s, over the budget) and says so when it skips | 60 s | 28 s for a one-component edit with a fresh `dist` (unit 0.4, components 4.6, screenshots 6.7, routes 16.6; load ~15) |
+| `bun run test` = `test:changed` | what the diff affects: `bun test --changed` (unit); the component tests whose imports reach a changed file (`scripts/test-graph.ts`: a static import graph with the barrel flattened, because Vitest's own `--changed` selects every file in browser mode); the screenshots of those components; the docs routes mapped by `scripts/route-impact.ts`. Prints what it selected and why. Every component test runs when a shared file or a test-runtime dependency changed (styles, tokens, theme, stores, primitives, a test setup file, `package.json`, `bun.lock`): the run prints `widened to all components because <file>` and the budget widens with it. The site part needs a fresh `site/dist` (a build is ~75 s, over the budget) and says so when it skips | 60 s narrow; 240 s (`landShared`) when widened to all components (`scripts/test-changed.sh` reads it from `--print-budget`) | 28 s for a one-component edit with a fresh `dist` (unit 0.4, components 4.6, screenshots 6.7, routes 16.6; load ~15). Widened by the happy-dom bump (#588): 95 s at load 10.5 for the full component set (~2190 tests), which is why the narrow 60 s does not apply to it |
 | `bun scripts/test-changed.ts --land` | the same plus the package and site builds and 5 canary routes (home, recipes, app-shell, templates/app, theme). This is the gate `issue-land.sh` runs; `validate` runs on the push. The package builds first (the unit tests that read `dist` need it), then the site build runs beside the unit and component tests (the component tests render from `src`). **When a shared file changed** (styles, tokens, theme, stores, primitives, a test setup file, the site layout, or a file the route map cannot place) it runs every component test and a **thinned route sweep** and **no screenshots**, under the larger budget; `--print-budget` says which one applies. The thinned sweep (`E2E_GALLERY=representatives`) keeps 6 gallery items, 3 atlas terms and the fewest docs pages that still render every component folder and subpath (`docsRepresentatives` in `scripts/route-impact.ts`: 81 of 160), plus every page the diff itself changed (`E2E_KEEP`); `test:full` audits all of them. The ~880 social cards are cached by the sha256 of their HTML and the browser version in `$TMPDIR/ca-og-cards` (shared by every worktree, `CA_OG_CACHE_DIR` overrides, cards unused for 30 days are pruned) and rendered four at a time | 180 s; 240 s when a shared file changed | 104 s for a one-component edit (builds 17 + 58, routes 17; load ~15). Shared case at load 7 (components ALL 210 files): 193.5 s with a warm card cache (package 11 + site build 94 beside the tests + routes 88) and 224.8 s with a cold one (site build 125); it was 382.5 s before #598 (unit 8 + components 80 + package 12 + site 134 + routes 148) |
 | `bun run validate` | themes, token lint, Biome, unit tests, both type checks, build, size | 90 s | 51 s at load 11 |
 | `bun run test:full` | unit, every component test, every route (layout, axe, dist-html, every project including WebKit), fixtures. Release gate, part 1 of 2 | 300 s | components 109 s (load 25) + routes 168 s (load 29) + unit 40 s + fixtures 7 s: over budget on a busy machine, not measured on a quiet one |

package.json

@@ -31,8 +31,8 @@
 		"lint:fix": "biome check --write .",
 		"validate": "bun scripts/test-run.ts --budget=validate bun scripts/validate.ts",
 		"postinstall": "node scripts/fix-ts-symlink.js && node scripts/install-hooks.js && bunx playwright install chromium webkit",
-		"test": "bun scripts/test-run.ts --budget=changed bun scripts/test-changed.ts",
-		"test:changed": "bun scripts/test-run.ts --budget=changed bun scripts/test-changed.ts",
+		"test": "sh scripts/test-changed.sh",
+		"test:changed": "sh scripts/test-changed.sh",
 		"test:full": "bun scripts/test-run.ts --budget=full sh -c 'bun run test:unit && bun run test:browser && bun run test:fixtures && bun run test:e2e'",
 		"test:unit": "bun run --cwd packages/ui test:unit && bun run --cwd site test:unit",
 		"test:browser": "bun run --cwd packages/ui test:browser",

packages/ui/tests/test-graph.test.ts

@@ -1,5 +1,11 @@
 import { describe, expect, it } from "bun:test";
-import { affected, buildGraph, isSharedComponent, landBudget } from "../../../scripts/test-graph";
+import {
+	affected,
+	buildGraph,
+	isSharedComponent,
+	resolveBudget,
+	widenedBecause,
+} from "../../../scripts/test-graph";
 
 // scripts/test-graph.ts selects component tests by a static import graph, because Vitest's `--changed`
 // selects every file in browser mode. Too wide a selection defeats it; too narrow hides a regression.
@@ -70,13 +76,21 @@ describe("test-graph", () => {
 	});
 });
 
-// `issue-land.sh` picks the land budget from this: a shared change runs every component test and every route
-// (360 s, no screenshots), anything else the affected ones plus the canaries (180 s).
-describe("landBudget", () => {
-	it("a component edit is under the 180 s budget", () => {
-		expect(landBudget([`${dir}/Avatar/Avatar.svelte`])).toBe("land");
-		expect(landBudget([`${dir}/Avatar/Avatar.svelte.test.ts`, "AGENTS.md"])).toBe("land");
-		expect(landBudget([])).toBe("land");
+// The budget follows the scope of the selection. A dependency bump (happy-dom, #588) correctly selects every
+// component test, and the full set alone is ~80 s (95 s at load 10.5), over the 60 s `changed` budget that
+// assumed a narrow selection. Widened to all components: `landShared`; narrow: `changed` (dev) or `land`.
+describe("resolveBudget", () => {
+	const narrow = [`${dir}/Avatar/Avatar.svelte`];
+
+	it("a narrow selection keeps its budget and has no widening reason", () => {
+		expect(resolveBudget(narrow, false)).toBe("changed");
+		expect(resolveBudget([`${dir}/Avatar/Avatar.svelte.test.ts`, "AGENTS.md"], false)).toBe(
+			"changed"
+		);
+		expect(resolveBudget([], false)).toBe("changed");
+		expect(resolveBudget(narrow, true)).toBe("land");
+		expect(resolveBudget([], true)).toBe("land");
+		expect(widenedBecause(narrow)).toBeUndefined();
 	});
 
 	it.each([
@@ -86,16 +100,18 @@ describe("landBudget", () => {
 		"packages/ui/src/lib/primitives/presence.svelte.ts",
 		"packages/ui/tests/setup.browser.ts",
 		"packages/ui/vite.config.ts",
+		"packages/ui/package.json",
 		"site/src/layouts/Base.astro",
-	])("%s is shared: the 360 s budget", (file) => {
-		expect(landBudget([file])).toBe("landShared");
+	])("%s widens to all components: landShared, dev run and land gate", (file) => {
+		for (const land of [false, true])
+			expect(resolveBudget([...narrow, file], land)).toBe("landShared");
+		expect(widenedBecause([...narrow, file])).toBe(file);
 	});
 
-	it("one shared file among ordinary ones is enough, and a file the route map does not know is shared", () => {
-		expect(landBudget([`${dir}/Avatar/Avatar.svelte`, "packages/ui/src/lib/styles/base.css"])).toBe(
-			"landShared"
-		);
-		expect(landBudget(["site/public/some-new-asset.svg"])).toBe("landShared");
+	it("a file the route map does not know widens too, and the reason names it", () => {
+		const file = "site/public/some-new-asset.svg";
+		expect(resolveBudget([file], false)).toBe("landShared");
+		expect(widenedBecause([...narrow, file])).toBe(file);
 	});
 
 	it("isSharedComponent only flags what every component test renders through", () => {

scripts/test-changed.sh

@@ -0,0 +1,9 @@
+#!/bin/sh
+# `bun run test` / `test:changed`: the budget follows the selection's scope. A narrow selection keeps the 60 s
+# `changed` budget; one widened to every component test (a shared file, a test-runtime dependency like happy-dom)
+# runs under `landShared`, and test-changed.ts says why. The budget is read before the clock starts.
+#   sh scripts/test-changed.sh [test-changed args]
+set -e
+cd "$(dirname "$0")/.."
+BUDGET=$(bun scripts/test-changed.ts --print-budget "$@")
+exec bun scripts/test-run.ts --budget="$BUDGET" bun scripts/test-changed.ts "$@"

scripts/test-changed.ts

@@ -8,7 +8,10 @@
  * A SHARED change (a shared component file, or a route-wide one: styles, tokens, theme, primitives, the
  * site layout, a test setup file, or a file the route map cannot place) runs every component test and, when
  * it reaches the site, every route, and NO screenshots: those run at release (`test:vrt`) and on demand.
- * `--print-budget` prints the budget this selection runs under (`land`, 180 s, or `landShared`, 240 s).
+ * `--print-budget` prints the budget this selection runs under: `changed` (60 s) for a narrow dev run, `land`
+ * (180 s) for a narrow land run, `landShared` (240 s) for either when the selection widens to every component
+ * test (a shared file or a test-runtime dependency such as happy-dom); the run then prints
+ * "widened to all components because <file>".
  *
  *   unit        `bun test --changed <base>` (module graph);
  *   components  the component tests whose imports reach a changed file (scripts/test-graph.ts; Vitest's
@@ -34,7 +37,14 @@ import { existsSync, readFileSync, statSync } from "node:fs";
 import { join } from "node:path";
 import { distStale } from "./dist-fresh.ts";
 import { lockReachesRuntime, ownRoutes, ROOT, select } from "./route-impact.ts";
-import { affected, buildGraph, isSharedComponent, landBudget } from "./test-graph.ts";
+import {
+	affected,
+	buildGraph,
+	isSharedComponent,
+	resolveBudget,
+	widenedBecause,
+} from "./test-graph.ts";
+import { BUDGETS } from "./test-run.ts";
 
 if (!import.meta.main) throw new Error("scripts/test-changed.ts is a command, not a module");
 
@@ -88,6 +98,13 @@ const changed = [
 	]),
 ].filter((f) => f !== "bun.lock" || !lockIsDevOnly());
 const short = base.slice(0, 8);
+// The budget follows the selection's scope (scripts/test-graph.ts `resolveBudget`): scripts/test-changed.sh and
+// `issue-land.sh` read it from here before the clock starts.
+const budget = resolveBudget(changed, land);
+if (process.argv.includes("--print-budget")) {
+	console.log(budget);
+	process.exit(0);
+}
 if (!changed.length && !land) {
 	console.log(`test:changed: no changes against origin/main (${short}), nothing to run`);
 	process.exit(0);
@@ -103,10 +120,11 @@ const visualRun = sel.all || sel.routes.some((r) => VISUAL_ROUTES.has(r));
 const routeRun = sweepAll || routes.length > 0;
 // Shared: everything that can reach every test runs, and screenshots are left to `test:vrt`.
 const shared = sel.all || sharedFile !== undefined;
-if (process.argv.includes("--print-budget")) {
-	console.log(land ? landBudget(changed) : "changed");
-	process.exit(0);
-}
+const widened = widenedBecause(changed);
+if (widened !== undefined)
+	console.log(
+		`test:changed: widened to all components because ${widened}, so the ${budget} budget applies (${BUDGETS[budget]} s)`
+	);
 
 console.log(
 	`test:changed${land ? " --land" : ""}: ${changed.length} changed file(s) against ${short}`

scripts/test-graph.ts

@@ -206,10 +206,22 @@ export const isSharedComponent = (file: string) =>
 	!isTestFile(file) && SHARED_COMPONENT.some((re) => re.test(file));
 
 /**
- * The budget `issue-land.sh` runs the land gate under: a SHARED change (a shared component file, or one the
- * route map says reaches every page or does not know) runs every component test and every route and gets
- * `landShared`; any other change runs what it affects plus the canaries under `land`.
+ * Why a selection widens to EVERY component test: the first shared file, or the file the route map says reaches
+ * every page or does not know. `undefined` when the selection is narrow.
  */
-export function landBudget(changed: string[]): "land" | "landShared" {
-	return changed.some(isSharedComponent) || select(changed).all ? "landShared" : "land";
+export function widenedBecause(changed: string[]): string | undefined {
+	const shared = changed.find(isSharedComponent);
+	if (shared !== undefined) return shared;
+	const sel = select(changed);
+	return sel.all ? sel.reason.split(":")[0] : undefined;
+}
+
+/**
+ * The budget a run gets, from the scope of its selection (the budget assumed a narrow one: a dependency bump
+ * selects every component test, ~80 s alone, 95 s at load 10.5 in #588). Widened to all components: `landShared`
+ * (dev run and land gate alike). Narrow: `changed` (60 s) for a dev run, `land` (180 s) with the canaries.
+ */
+export function resolveBudget(changed: string[], land: boolean): "changed" | "land" | "landShared" {
+	if (widenedBecause(changed) !== undefined) return "landShared";
+	return land ? "land" : "changed";
 }

scripts/test-run.ts

@@ -194,7 +194,7 @@ export function releaseLock(pid: number) {
 
 /** Wall-time budget in seconds of each command. Raise one only in its own commit, with the measured reason. */
 export const BUDGETS = {
-	/** `bun run test` / `test:changed`: what the diff against origin/main affects */
+	/** `bun run test` / `test:changed` with a NARROW selection (one that widens to every component test runs under `landShared`, scripts/test-graph.ts `resolveBudget`) */
 	changed: 60,
 	/** `bun run validate`: build, types, lint, unit tests, size */
 	validate: 90,
@@ -209,6 +209,8 @@ export const BUDGETS = {
 	 * 1044 (#598). Measured at load 7 on 12 cores, a shared change (components ALL 210 files): 224.8 s with a cold card
 	 * cache (package 11 s, then site build 125 s beside the tests, then routes 88 s) and 193.5 s with a warm one (site
 	 * build 94 s). The full sweep stays at release (`test:full`). Do not raise it again: make the run faster.
+	 * A dev run (`bun run test`) widened to every component test gets it too: the full component set alone is ~80 s
+	 * (95 s at load 10.5 on the happy-dom bump #588), over the 60 s `changed` budget that assumed a narrow selection.
 	 */
 	landShared: 240,
 	/** `bun run test:full`: unit, every component test, every route, fixtures (release gate, part 1) */