pcbjam/tests/tools/screenshots/post-discord.ts

335 lines
14 KiB
TypeScript
Raw Normal View History

feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
/**
* The always-on CI-on-main Discord report.
*
* One message leads with the commit SHA + e2e pass/fail + the runtime-perf table
* (perf-report.ts), then screenshot changes are attached as triptychs
* (old | new+boxes | heatmap), ADDED images, and a REMOVED list. Because perf is
* always present, the comment fires on every main commit even with no screenshot
* change — "see it in hindsight on every commit."
*
* Safety / robustness:
* - Posts only on push to main (unless --force); a missing DISCORD_WEBHOOK_URL is
* a silent no-op, so it's inert on PRs / forks (which can't see the secret).
* - Attachments batched ≤10/message and downscaled to fit Discord's size caps;
* 429 Retry-After honoured; first-run flood (hundreds of ADDED) collapsed.
* - --dry-run composes everything and prints it without POSTing (how it's tested).
*
* CLI (from tests/):
* tsx tools/screenshots/post-discord.ts [--dry-run] [--force] [--e2e pass|fail] [--subject S]
*/
import * as fs from 'fs';
import * as path from 'path';
import { execFileSync } from 'child_process';
e2e/CI: dual-engine suites, per-engine screenshots, SwiftShader retired, prod web suite, CI-coverage gate Squash of experiment/ff-big-modules vs main. Big-module routing removed: native-EH shrank kicad_editor below SpiderMonkey's x86-64 code budget (runs 29355049705/29356152413 green on stock Firefox), so BIG_MODULE_SPECS routing and the baseline-only-JIT crutch are gone — kicad-firefox and kicad-chromium both run the full suite, with the module compiled the way real users' browsers compile it. Per-engine screenshots end to end: stableShot/shotPath write test-results/<engine>/<name>.png; baselines move to baseline-screenshots/{chromium,firefox}/ and the whole tools/screenshots pipeline (compare/promote/manifest/spec-map/changelog/Discord) keys on <engine>/<name>. Previously Firefox and Chromium renders of one spec overwrote each other and Firefox renders were never actually gated. Seeded from CI run 29421380806 (92 new firefox baselines, +24 chromium web-suite shots); manifest generated from the baseline tree. One merged playwright.config.ts (kicad/asyncify/coroutine/perf as projects); ~25 dead npm scripts dropped. The web suite is gated in CI for the first time ever (4 rotted specs fixed, 5 broken lib-bridge specs triaged as fixme in docs/features/web-e2e-rot/); cheap lint step after npm ci; last 26 blind-sleep violations fixed. SwiftShader retired: CI Chromium renders WebGL on ANGLE → Mesa llvmpipe (--use-gl=angle --use-angle=gl --ignore-gpu-blocklist; the blocklist flag is mandatory — llvmpipe is blocklisted and WebGL is silently unavailable without it) in BOTH configs. Under WORKERS=4 congestion SwiftShader transiently failed the first post-board-load draw and the recovery cascade ended in a silent permanent Cairo fallback — that engine flip was the "~1.2% changedRatio both directions" occ-export baseline flake. Validated 160/160 across two 80-repeat rigs; full analysis in docs/features/wx-parity-bugs/occ-export-context-eviction.md. Chromium baselines shift slightly on llvmpipe — promote once from the first green run. Deflakes the new coverage exposed: presence baselines settle before capture; presence fixtures declare current file formats; perf gets its own outputDir so CI evidence survives; occ-export settles the board paint before the export dialog; menu-item waits (waitForRenderedByLabel before clickMenuItem) in 4 specs + the TESTING.md rule. Web suite runs the PROD build, in parallel: webServer becomes backend `start` + the standalone's e2e:preview (build-preview.mjs: link-wasm → stash the public/wasm symlink aside during vite build, build-demo.mjs's move — then vite preview as the persistent server). The wasm middleware serves /wasm/* in preview and emits COOP/COEP/CORP itself (a pthread worker script's own response must carry COEP or Chrome kills it with ERR_BLOCKED_BY_RESPONSE). VITE_* flags bake at build time; VITE_ALLOW_USER_OVERRIDE joins turbo globalEnv. fullyParallel + default workers: 5.2m → 1.4m. Determinism fixes the parallel run exposed: shared-page specs become serial groups; locks.spec grabs alice's exact item via the new kicadCollabTestSelectByUuid hook (cross-tab "first footprint" order is not a ysync invariant); quit specs poll page.url() (quit supersedes its own navigation — NS_BINDING_ABORTED on Firefox). Suite: 51 passed / 12 skipped / 0 failed in 1.6m. CI-coverage gate (lint:ci-coverage): every tests/**/*.spec.ts must be reachable from the npm scripts the workflows invoke — scraped from .github/workflows/, resolved through package.json, coverage asked from playwright --list itself. Rules: uncovered-spec + orphan-project (with a documented LOCAL_ONLY_PROJECTS allowlist). Gating next to lint:determinism; 138 spec files / 13 projects accounted for. Product fixes kept from the investigations (reachable on real GPUs too): wx 7799fd1be5 — paint flags clear before dispatch + Invalidate always propagates; kicad 3dcfea5e45 — SwiftShader pass-boundary flush + per-instance font texture + first-frame GL-error drain (GAL recovery recovers instead of falling back to Cairo) + the user-facing eeschema switch navigates again under __EMSCRIPTEN__ (project-sync's FaceRegistered gate had rerouted it into the hidden sync player; caught by the newly-gated web suite). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018eUxiPApHgGiu9NFyQfhAq
2026-07-17 12:10:40 +02:00
import { DIFF_OUT_DIR, splitKey } from './config';
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
import { loadPng, encodeWithinCap } from './image-ops';
import { buildPerfReport, fetchPreviousPerf } from './perf-report';
import type { Report } from './compare';
const MAX_FILE_BYTES = 8 * 1024 * 1024;
const MAX_MSG_BYTES = 24 * 1024 * 1024;
const MAX_FILES_PER_MSG = 9; // ≤9 images per Discord message; the rest spill into more messages
const MAX_CONTENT = 1000; // split message text into multiple messages beyond this many chars
const MAX_TOTAL_FILES = 30; // ~4 messages of images; excess is summarized, not posted
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
const FLOOD_N = 12; // more ADDED than this ⇒ show a few exemplars, not all
const REMOVED_CAP = 40; // cap the REMOVED name list in the message body
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
type Attachment = { name: string; buffer: Buffer };
type Message = { content: string; files: Attachment[] };
function sanitize(name: string): string {
return name.replace(/[^\w.-]+/g, '_');
}
function e2eBadge(status: string | undefined): string {
if (status === 'pass') return '✅ e2e passed';
if (status === 'fail') return '❌ e2e failed';
return 'ℹ️ e2e status unknown';
}
function commitSubject(sha: string | undefined): string {
if (!sha) return '';
try {
return execFileSync('git', ['log', '-1', '--pretty=%s', sha], { encoding: 'utf8' }).trim();
} catch {
return '';
}
}
function readReport(root: string): Report | null {
const p = path.join(root, DIFF_OUT_DIR, 'report.json');
if (!fs.existsSync(p)) return null;
try {
return JSON.parse(fs.readFileSync(p, 'utf8')) as Report;
} catch {
return null;
}
}
/** Encode a PNG path to a within-cap attachment, or null if missing/unreadable. */
function attach(root: string, rel: string, name: string): Attachment | null {
const abs = path.isAbsolute(rel) ? rel : path.join(root, rel);
if (!fs.existsSync(abs)) return null;
try {
return { name: sanitize(name), buffer: encodeWithinCap(loadPng(abs), MAX_FILE_BYTES) };
} catch {
return null;
}
}
/** One status line for the 3D-renderer regression suite (tests/3d-regression),
* from the compare-dirs report.json of each level that ran. Empty when the
* suite didn't run (reports absent) so the message stays inert for builds
* without the 3D harness. Triptychs live in the run artifact, not Discord. */
function threeDStatusLine(): string[] {
const levels: Array<{ tag: string; dir: string }> = [
{ tag: 'parity', dir: '3d-regression/output/diff/parity' },
{ tag: 'webgl-self', dir: '3d-regression/output/diff/webgl-self' },
];
const parts: string[] = [];
let drift = false;
for (const { tag, dir } of levels) {
const file = path.join(dir, 'report.json');
if (!fs.existsSync(file)) continue;
try {
const r = JSON.parse(fs.readFileSync(file, 'utf8'));
const changed = (r.changed?.length ?? 0) + (r.added?.length ?? 0) + (r.removed?.length ?? 0);
const total = (r.unchangedCount ?? 0) + (r.changed?.length ?? 0);
if (changed === 0) {
parts.push(`${tag} ${total}/${total}`);
} else {
drift = true;
const worst = r.changed?.[0];
const worstNote = worst ? ` (worst \`${worst.name}\` ${(worst.changedRatio * 100).toFixed(1)}%)` : '';
parts.push(`**${tag} ${changed} changed**${worstNote}`);
}
} catch {
/* unreadable report — skip the level */
}
}
if (!parts.length) return [];
return [`${drift ? '⚠️' : '✅'} 3D renderer: ${parts.join(' · ')}${drift ? ' — triptychs in the run artifact' : ''}`];
}
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
/** Build the header text: SHA, subject, e2e, perf table, screenshot summary, removed list. */
function buildHeader(report: Report | null, perfBlock: string, meta: { sha?: string; subject: string; e2e?: string }): string {
const shortSha = meta.sha ? meta.sha.slice(0, 7) : 'local';
const lines = [`📸 **CI screenshot + perf report** · \`${shortSha}\``];
if (meta.subject) lines.push(`> ${meta.subject}`);
lines.push(e2eBadge(meta.e2e));
if (perfBlock) lines.push('', perfBlock);
e2e/CI: dual-engine suites, per-engine screenshots, SwiftShader retired, prod web suite, CI-coverage gate Squash of experiment/ff-big-modules vs main. Big-module routing removed: native-EH shrank kicad_editor below SpiderMonkey's x86-64 code budget (runs 29355049705/29356152413 green on stock Firefox), so BIG_MODULE_SPECS routing and the baseline-only-JIT crutch are gone — kicad-firefox and kicad-chromium both run the full suite, with the module compiled the way real users' browsers compile it. Per-engine screenshots end to end: stableShot/shotPath write test-results/<engine>/<name>.png; baselines move to baseline-screenshots/{chromium,firefox}/ and the whole tools/screenshots pipeline (compare/promote/manifest/spec-map/changelog/Discord) keys on <engine>/<name>. Previously Firefox and Chromium renders of one spec overwrote each other and Firefox renders were never actually gated. Seeded from CI run 29421380806 (92 new firefox baselines, +24 chromium web-suite shots); manifest generated from the baseline tree. One merged playwright.config.ts (kicad/asyncify/coroutine/perf as projects); ~25 dead npm scripts dropped. The web suite is gated in CI for the first time ever (4 rotted specs fixed, 5 broken lib-bridge specs triaged as fixme in docs/features/web-e2e-rot/); cheap lint step after npm ci; last 26 blind-sleep violations fixed. SwiftShader retired: CI Chromium renders WebGL on ANGLE → Mesa llvmpipe (--use-gl=angle --use-angle=gl --ignore-gpu-blocklist; the blocklist flag is mandatory — llvmpipe is blocklisted and WebGL is silently unavailable without it) in BOTH configs. Under WORKERS=4 congestion SwiftShader transiently failed the first post-board-load draw and the recovery cascade ended in a silent permanent Cairo fallback — that engine flip was the "~1.2% changedRatio both directions" occ-export baseline flake. Validated 160/160 across two 80-repeat rigs; full analysis in docs/features/wx-parity-bugs/occ-export-context-eviction.md. Chromium baselines shift slightly on llvmpipe — promote once from the first green run. Deflakes the new coverage exposed: presence baselines settle before capture; presence fixtures declare current file formats; perf gets its own outputDir so CI evidence survives; occ-export settles the board paint before the export dialog; menu-item waits (waitForRenderedByLabel before clickMenuItem) in 4 specs + the TESTING.md rule. Web suite runs the PROD build, in parallel: webServer becomes backend `start` + the standalone's e2e:preview (build-preview.mjs: link-wasm → stash the public/wasm symlink aside during vite build, build-demo.mjs's move — then vite preview as the persistent server). The wasm middleware serves /wasm/* in preview and emits COOP/COEP/CORP itself (a pthread worker script's own response must carry COEP or Chrome kills it with ERR_BLOCKED_BY_RESPONSE). VITE_* flags bake at build time; VITE_ALLOW_USER_OVERRIDE joins turbo globalEnv. fullyParallel + default workers: 5.2m → 1.4m. Determinism fixes the parallel run exposed: shared-page specs become serial groups; locks.spec grabs alice's exact item via the new kicadCollabTestSelectByUuid hook (cross-tab "first footprint" order is not a ysync invariant); quit specs poll page.url() (quit supersedes its own navigation — NS_BINDING_ABORTED on Firefox). Suite: 51 passed / 12 skipped / 0 failed in 1.6m. CI-coverage gate (lint:ci-coverage): every tests/**/*.spec.ts must be reachable from the npm scripts the workflows invoke — scraped from .github/workflows/, resolved through package.json, coverage asked from playwright --list itself. Rules: uncovered-spec + orphan-project (with a documented LOCAL_ONLY_PROJECTS allowlist). Gating next to lint:determinism; 138 spec files / 13 projects accounted for. Product fixes kept from the investigations (reachable on real GPUs too): wx 7799fd1be5 — paint flags clear before dispatch + Invalidate always propagates; kicad 3dcfea5e45 — SwiftShader pass-boundary flush + per-instance font texture + first-frame GL-error drain (GAL recovery recovers instead of falling back to Cairo) + the user-facing eeschema switch navigates again under __EMSCRIPTEN__ (project-sync's FaceRegistered gate had rerouted it into the hidden sync player; caught by the newly-gated web suite). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018eUxiPApHgGiu9NFyQfhAq
2026-07-17 12:10:40 +02:00
// Per-engine breakdown suffix, e.g. " (2 chromium / 1 firefox)"; single-engine
// sets show just the engine name so the origin is always visible.
const byEngine = (entries: Array<{ name: string }>): string => {
if (!entries.length) return '';
const counts = new Map<string, number>();
for (const e of entries) {
const eng = splitKey(e.name).engine;
counts.set(eng, (counts.get(eng) ?? 0) + 1);
}
if (counts.size < 2) return ` (${[...counts.keys()][0]})`;
return ` (${[...counts.entries()].map(([e, n]) => `${n} ${e}`).join(' / ')})`;
};
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
lines.push('');
if (!report || (!report.changed.length && !report.added.length && !report.removed.length)) {
lines.push('✅ no screenshot drift');
} else if (report.driftLikely) {
lines.push(`⚠️ **${report.changed.length} screenshots changed broadly** — looks like host env drift → re-promote (\`npm run screenshots:promote\`)`);
} else {
e2e/CI: dual-engine suites, per-engine screenshots, SwiftShader retired, prod web suite, CI-coverage gate Squash of experiment/ff-big-modules vs main. Big-module routing removed: native-EH shrank kicad_editor below SpiderMonkey's x86-64 code budget (runs 29355049705/29356152413 green on stock Firefox), so BIG_MODULE_SPECS routing and the baseline-only-JIT crutch are gone — kicad-firefox and kicad-chromium both run the full suite, with the module compiled the way real users' browsers compile it. Per-engine screenshots end to end: stableShot/shotPath write test-results/<engine>/<name>.png; baselines move to baseline-screenshots/{chromium,firefox}/ and the whole tools/screenshots pipeline (compare/promote/manifest/spec-map/changelog/Discord) keys on <engine>/<name>. Previously Firefox and Chromium renders of one spec overwrote each other and Firefox renders were never actually gated. Seeded from CI run 29421380806 (92 new firefox baselines, +24 chromium web-suite shots); manifest generated from the baseline tree. One merged playwright.config.ts (kicad/asyncify/coroutine/perf as projects); ~25 dead npm scripts dropped. The web suite is gated in CI for the first time ever (4 rotted specs fixed, 5 broken lib-bridge specs triaged as fixme in docs/features/web-e2e-rot/); cheap lint step after npm ci; last 26 blind-sleep violations fixed. SwiftShader retired: CI Chromium renders WebGL on ANGLE → Mesa llvmpipe (--use-gl=angle --use-angle=gl --ignore-gpu-blocklist; the blocklist flag is mandatory — llvmpipe is blocklisted and WebGL is silently unavailable without it) in BOTH configs. Under WORKERS=4 congestion SwiftShader transiently failed the first post-board-load draw and the recovery cascade ended in a silent permanent Cairo fallback — that engine flip was the "~1.2% changedRatio both directions" occ-export baseline flake. Validated 160/160 across two 80-repeat rigs; full analysis in docs/features/wx-parity-bugs/occ-export-context-eviction.md. Chromium baselines shift slightly on llvmpipe — promote once from the first green run. Deflakes the new coverage exposed: presence baselines settle before capture; presence fixtures declare current file formats; perf gets its own outputDir so CI evidence survives; occ-export settles the board paint before the export dialog; menu-item waits (waitForRenderedByLabel before clickMenuItem) in 4 specs + the TESTING.md rule. Web suite runs the PROD build, in parallel: webServer becomes backend `start` + the standalone's e2e:preview (build-preview.mjs: link-wasm → stash the public/wasm symlink aside during vite build, build-demo.mjs's move — then vite preview as the persistent server). The wasm middleware serves /wasm/* in preview and emits COOP/COEP/CORP itself (a pthread worker script's own response must carry COEP or Chrome kills it with ERR_BLOCKED_BY_RESPONSE). VITE_* flags bake at build time; VITE_ALLOW_USER_OVERRIDE joins turbo globalEnv. fullyParallel + default workers: 5.2m → 1.4m. Determinism fixes the parallel run exposed: shared-page specs become serial groups; locks.spec grabs alice's exact item via the new kicadCollabTestSelectByUuid hook (cross-tab "first footprint" order is not a ysync invariant); quit specs poll page.url() (quit supersedes its own navigation — NS_BINDING_ABORTED on Firefox). Suite: 51 passed / 12 skipped / 0 failed in 1.6m. CI-coverage gate (lint:ci-coverage): every tests/**/*.spec.ts must be reachable from the npm scripts the workflows invoke — scraped from .github/workflows/, resolved through package.json, coverage asked from playwright --list itself. Rules: uncovered-spec + orphan-project (with a documented LOCAL_ONLY_PROJECTS allowlist). Gating next to lint:determinism; 138 spec files / 13 projects accounted for. Product fixes kept from the investigations (reachable on real GPUs too): wx 7799fd1be5 — paint flags clear before dispatch + Invalidate always propagates; kicad 3dcfea5e45 — SwiftShader pass-boundary flush + per-instance font texture + first-frame GL-error drain (GAL recovery recovers instead of falling back to Cairo) + the user-facing eeschema switch navigates again under __EMSCRIPTEN__ (project-sync's FaceRegistered gate had rerouted it into the hidden sync player; caught by the newly-gated web suite). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_018eUxiPApHgGiu9NFyQfhAq
2026-07-17 12:10:40 +02:00
lines.push(
`⚠️ **screenshot drift**: ${report.changed.length} changed${byEngine(report.changed)}, ` +
`${report.added.length} added${byEngine(report.added)}, ` +
`${report.removed.length} removed${byEngine(report.removed)}`
);
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
}
if (report?.removed.length) {
const shown = report.removed.slice(0, REMOVED_CAP).map((r) => `\`${r.name}\``).join(', ');
const extra = report.removed.length > REMOVED_CAP ? ` (+${report.removed.length - REMOVED_CAP} more)` : '';
lines.push('➖ REMOVED: ' + shown + extra);
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
}
lines.push(...threeDStatusLine());
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
return lines.join('\n');
}
/** Collect the image attachments (changed triptychs + added images) with flood-collapse + total cap. */
export function buildAttachments(root: string, report: Report | null): { files: Attachment[]; notes: string[] } {
const files: Attachment[] = [];
const notes: string[] = [];
if (!report) return { files, notes };
for (const c of report.changed) {
if (files.length >= MAX_TOTAL_FILES) break;
const a = attach(root, c.triptych, `CHANGED_${c.name}`);
if (a) files.push(a);
}
const addedToShow = report.added.length > FLOOD_N ? report.added.slice(0, 3) : report.added;
if (report.added.length > FLOOD_N) {
notes.push(`➕ ${report.added.length} added (showing ${addedToShow.length})`);
}
for (const ad of addedToShow) {
if (files.length >= MAX_TOTAL_FILES) break;
const a = attach(root, ad.image, `ADDED_${ad.name}`);
if (a) files.push(a);
}
const removedToShow = report.removed.length > FLOOD_N ? report.removed.slice(0, 3) : report.removed;
if (report.removed.length > FLOOD_N) {
notes.push(`➖ ${report.removed.length} removed (showing ${removedToShow.length})`);
}
for (const rm of removedToShow) {
if (files.length >= MAX_TOTAL_FILES) break;
const a = attach(root, rm.image, `REMOVED_${rm.name}`);
if (a) files.push(a);
}
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
const shown = files.length;
const wanted = report.changed.length + addedToShow.length + removedToShow.length;
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
if (wanted > shown) notes.push(`(${wanted - shown} more images omitted — see the CI artifact)`);
return { files, notes };
}
/**
* Assemble Discord messages: first the text content split into ≤MAX_CONTENT-char
* messages (at line boundaries, never inside a ``` fence so the perf table stays
* intact), then the image attachments in messages of ≤MAX_FILES_PER_MSG files.
*/
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
export function paginate(header: string, files: Attachment[]): Message[] {
const messages: Message[] = [];
// 1) Text content → one or more text-only messages.
let cur = '';
let inFence = false;
for (const line of header.split('\n')) {
const next = cur ? cur + '\n' + line : line;
if (!inFence && cur && next.length > MAX_CONTENT) {
messages.push({ content: cur, files: [] });
cur = line;
} else {
cur = next;
}
if (line.trim().startsWith('```')) inFence = !inFence;
}
if (cur) messages.push({ content: cur, files: [] });
// 2) Image attachments → messages of ≤MAX_FILES_PER_MSG files and ≤MAX_MSG_BYTES.
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
let batch: Attachment[] = [];
let bytes = 0;
const flush = () => {
if (batch.length) messages.push({ content: '', files: batch });
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
batch = [];
bytes = 0;
};
for (const f of files) {
if (batch.length >= MAX_FILES_PER_MSG || bytes + f.buffer.length > MAX_MSG_BYTES) flush();
batch.push(f);
bytes += f.buffer.length;
}
flush();
return messages.length ? messages : [{ content: header, files: [] }];
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
}
async function sleep(ms: number): Promise<void> {
return new Promise((r) => setTimeout(r, ms));
}
export async function postMessage(webhook: string, msg: Message): Promise<void> {
for (let attemptNo = 0; attemptNo < 6; attemptNo++) {
const form = new FormData();
form.append('payload_json', JSON.stringify({ content: msg.content || '', allowed_mentions: { parse: [] } }));
msg.files.forEach((f, i) => form.append(`files[${i}]`, new Blob([new Uint8Array(f.buffer)], { type: 'image/png' }), f.name));
const res = await fetch(webhook, { method: 'POST', body: form });
if (res.ok) return;
if (res.status === 429) {
const retryAfter = Number(res.headers.get('retry-after')) || 2;
await sleep((retryAfter + 0.5) * 1000);
continue;
}
throw new Error(`Discord POST failed ${res.status}: ${(await res.text()).slice(0, 300)}`);
}
throw new Error('Discord POST failed after retries (429)');
}
function parseArgs(argv: string[]): Record<string, string | boolean> {
const out: Record<string, string | boolean> = {};
for (let i = 0; i < argv.length; i++) {
const a = argv[i];
if (a === '--dry-run') out.dryRun = true;
else if (a === '--force') out.force = true;
else if (a === '--e2e') out.e2e = argv[++i];
else if (a === '--subject') out.subject = argv[++i];
else if (a === '--repo') out.repo = argv[++i];
}
return out;
}
async function main(): Promise<void> {
const args = parseArgs(process.argv.slice(2));
const root = process.cwd();
const isMainPush = process.env.GITHUB_REF === 'refs/heads/main' && process.env.GITHUB_EVENT_NAME === 'push';
if (!args.force && !args.dryRun && !isMainPush) {
console.log('[discord] not a push to main — skipping');
return;
}
const webhook = process.env.DISCORD_WEBHOOK_URL;
if (!webhook && !args.dryRun) {
console.log('[discord] DISCORD_WEBHOOK_URL unset — skipping (inert on PRs/forks)');
return;
}
const sha = process.env.GITHUB_SHA;
const report = readReport(root);
// Loud pipeline alert: if the run produced ZERO screenshots (every baseline shows as
// "removed"), that's almost always a build/pipeline failure — not a real mass removal.
// Post a distinct, image-less alert instead of dumping a giant REMOVED list.
if (report && report.changed.length + report.added.length + report.unchangedCount === 0) {
const shortSha = sha ? sha.slice(0, 7) : 'local';
const runUrl =
process.env.GITHUB_SERVER_URL && process.env.GITHUB_REPOSITORY && process.env.GITHUB_RUN_ID
? `${process.env.GITHUB_SERVER_URL}/${process.env.GITHUB_REPOSITORY}/actions/runs/${process.env.GITHUB_RUN_ID}`
: '';
const alert =
`⚠️ **No screenshots produced this run** · \`${shortSha}\` — compare saw 0 rendered ` +
`vs ${report.removed.length} baselines. Likely a build/pipeline failure, not a mass removal.` +
(runUrl ? `\n${runUrl}` : '');
if (args.dryRun) {
console.log('--- message 1/1 (0 files) ---\n' + alert);
} else {
await postMessage(webhook!, { content: alert, files: [] });
console.log('[discord] posted pipeline alert (0 screenshots rendered)');
}
return;
}
feat(tests): screenshot regression + Discord review, perf-tracked New tooling in tests/tools/screenshots/ (TypeScript via tsx): - compare.ts: one pixelmatch engine (AA-excluded), connected-component "where to look" boxes, old|new+boxes|heatmap triptych, per-engine floors. - promote.ts: churn-free updater — overwrite a baseline only when decoded pixels differ beyond the floor, copying CI bytes verbatim (no re-encode churn); pulls a CI run via `gh run download` or a local --from dir. - post-discord.ts: always-on CI-on-main report (SHA + e2e status + the track-only runtime-perf table), then screenshot triptychs, batched + size-capped + flood-collapsed + 429-aware. - perf-report.ts: perf table with Δ vs the previous main run (via gh). - changelog.ts: no-build git-history baseline differ (Discord trigger B). - noise.ts / gen-manifest.ts: calibration + manifest generation. CI wiring: - wasm-build.yml: post-test step runs the gate + report on the already- produced test-results (no extra build); report-only (continue-on-error), posts only on push to main, inert without DISCORD_WEBHOOK_URL. - ci-ubicloud.yml: secrets: inherit (pass the webhook through). - screenshot-changelog.yml: ~30s no-build changelog on baseline changes. screenshot-manifest.json: canonical 354-name set + best-effort engine tags (313 chromium-swiftshader / 41 firefox-llvmpipe). Normalize scale:'device'->'css' across 18 spec files (no-op at CI DSF=1) so committed baselines are uniformly css-scaled. Design: CI's Linux render is the single source of truth; no pinned container (accept rare env drift -> re-promote); dev commits via promote. Replaces the byte-cmp compare-screenshots.sh + file-size-proxy update-baseline-screenshots.sh (kept for now until the first re-baseline). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-01 18:01:15 +02:00
const prevDir = fetchPreviousPerf((args.repo as string) || process.env.GITHUB_REPOSITORY, sha);
const { block: perfBlock } = buildPerfReport({ prevDir });
let header = buildHeader(report, perfBlock, {
sha,
subject: (args.subject as string) ?? commitSubject(sha),
e2e: args.e2e as string,
});
const { files, notes } = buildAttachments(root, report);
if (notes.length) header += '\n' + notes.join('\n');
const messages = paginate(header, files);
if (args.dryRun) {
for (const [i, m] of messages.entries()) {
console.log(`--- message ${i + 1}/${messages.length} (${m.files.length} files, ${m.files.reduce((s, f) => s + f.buffer.length, 0)} bytes) ---`);
if (m.content) console.log(m.content);
for (const f of m.files) console.log(` [attach] ${f.name} (${f.buffer.length} bytes)`);
}
return;
}
for (const m of messages) await postMessage(webhook!, m);
console.log(`[discord] posted ${messages.length} message(s)`);
}
if (require.main === module) {
main().catch((e) => {
console.error(`[discord] ${e.message}`);
process.exitCode = 1;
});
}