Skip to content

log.md

Latest commit

 

History

History
200 lines (150 loc) · 23.4 KB

log.md

File metadata and controls

200 lines (150 loc) · 23.4 KB
Raw

Engineering Log

This is an append-only record of repository changes and their verification. Add a new dated entry for each material change; do not rewrite earlier entries. Record only commands and outcomes that were actually observed.

2026-10-15 — Evidence workspace visual redesign

  • Reworked the application shell, primary navigation, test index, first-run overview, run selection, and results summary into a restrained evidence workspace without changing registry, execution, scoring, report, or provider behavior.
  • Made the readable light theme the default while preserving the existing working theme toggle; safety orange is reserved for run actions and keyboard focus, and result colors remain semantic.
  • Added docs/design-rationale.md with the UI audit, design read, decision reasons, and explicit draft-without-direction status.
  • Validation: npm run typecheck and npm run build passed; git diff --check completed without diff errors. Contrast checks passed for body text, muted text, the orange action color, and dark-mode foreground pairs.
  • Browser validation: a user-owned Vite server already held port 8080. Browser automation timed out while attaching to that session, so no rendered local UI claim is recorded for this change.
  • Deployment status: no deployment was performed. Current production remains stale and provider verification cannot be resolved from the local source until the current build is deployed to the exact verified host.

2026-10-15 — Integration readiness

  • Verified the complete feature/simplified-first-run branch before integration.
  • Validation: npm run typecheck, npm run build, and git diff --check passed.
  • Branch inventory: only main and feature/simplified-first-run existed locally/remotely; no additional worktrees or unrelated branch histories required merging.
  • Integration: merged into main as 03c77a5 Merge feature/simplified-first-run and pushed to origin/main.

2026-10-15 — Approved provider metadata and loader

  • Added the Monetag ownership-verification meta tag to index.html.
  • Added the operator-approved asynchronous Google AdSense loader to index.html.
  • Documented that the loader is third-party/exploratory and excluded from scoring, privacy-signal, and attribution results.
  • Validation: npm run build passed; the generated dist/index.html was confirmed to contain the AdSense loader.
  • Git commit: 24415c5 Add approved AdSense loader on feature/simplified-first-run.
  • Deployment status: not pushed, merged, or deployed. The deployed https://lab.frpboy.in/ page was observed without the Monetag verification meta tag before this local change is deployed.

2026-10-15 — First-run UX simplification

  • Added an Overview-first experience with primary navigation for Overview, Run a Test, Results, Learn & Methodology, and Settings.
  • Added Quick Scan using real registry cases DSP-01, TPA-01, TPA-05, TRK-01, INT-03, and AAB-01.
  • Full Diagnostic uses the full 55-case TEST_REGISTRY; both routes use the existing batch runner and route to Results.
  • Kept catalog, analysis/exports, score configuration, Monetag chamber, audit suite, and diagnostics reachable under Advanced Tools.
  • Validation: npm run typecheck and npm run build passed; Chromium local validation rendered Overview, completed Quick Scan, routed to Results, and showed no failed in-app audit assertions.
  • Git commits: 5a000c1 Refactor app navigation and enhance first-run experience; 196ac5b Document simplified first-run flow.

2026-10-15 — v4.1 registry integrity repair

  • Integrated the v4.1 lab implementation and reconciled the registry/UI/verification counts to 55 total cases: 35 core ad/tracker-blocking cases, 9 privacy signals, and 11 exploratory probes.
  • Kept privacy signals and exploratory probes outside the authoritative score.
  • Validation: typecheck, production build, dependency audit (0 vulnerabilities), Chromium smoke test, and all 15 in-app assertions passed.
  • Git commit: d5d1e15 Update lab to v4.1 implementation.

Entry template

YYYY-MM-DD — Short change title

  • Change:
  • Files:
  • Verification:
  • Deployment status:
  • Follow-up / known limitation:

2026-10-15 — Evidence workspace browser follow-up

  • Local Vite validation at http://127.0.0.1:8080 rendered the redesigned Overview with the 55-case index and all six Quick Scan IDs.
  • Verified primary navigation from Overview to Methodology and back, plus light and dark theme rendering.
  • No diagnostic run was started during this UI validation; registry, result, and score evidence remain untouched.
  • Deployment status: no deployment or provider re-verification was performed.

2026-10-15 — Mobile navigation correction

  • Mobile review found that the advanced controls and 55-cell index consumed too much first-viewport space.
  • Replaced the always-visible mobile advanced controls with a labelled More tools disclosure and replaced the mobile grid with a labelled route to the complete catalog.
  • Validation: npm run typecheck, npm run build, and git diff --check passed. Browser inspection at a 390px viewport verified both the collapsed state and expanded More tools state.
  • Updated the document theme color to match the default light evidence workspace.
  • Deployment status: no deployment was performed.

2026-10-15 — Production provider verification follow-up

  • Direct production inspection confirmed https://lab.frpboy.in/ serves the v4.1 evidence workspace, the Monetag verification meta tag, and the approved AdSense loader.
  • Direct Monetag dashboard inspection now marks lab.frpboy.in as Verified.
  • The dashboard still reports a zone-installation error for the configured live zone. No zone script was added because the repository policy requires explicit provider approval and a documented isolated exploratory-mode design before mounting live-ad code.

2026-10-15 — Browser validation, module extraction, and local fixture protocol

  • Baseline: local main and origin/main both resolved to 3c4ef61315edddcff727548576fdd9efa26a79c2 before this change.
  • Production validation: https://lab.frpboy.in/ rendered v4.1.0 at 1920x873 and showed the six Quick Scan IDs. A 390x844 capture showed labelled More tools, Browse all 55 test records, and no horizontal overflow (scrollWidth=375, clientWidth=375). Captured Chrome user agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/155.0.0.0 Safari/537.36. Desktop and 390px screenshots were captured through browser automation during this validation.
  • Deployment evidence: the Vercel production deployment 3b3FBXfy1PuRAsnz3RLWjj4r6RoG was Ready and identified source commit 8dd638df37f78ec28ea71c5f9afdd26f4a94dc22; it is behind local/current main at 3c4ef61. Production therefore cannot be claimed to be deployed from latest main.
  • Local interaction evidence at http://127.0.0.1:8080: empty Results view showed an unavailable score with zero completed results; Quick Scan completed six cases and showed a 0/100 score from 6/35 eligible results in this Chrome profile; DSP-01 detail showed all six trace stages. The Full Diagnostic run reached 51 completed cases during observation. Pause/cancel controls and escaped-HTML download could not be conclusively exercised because the run completed before the control action was delivered and the browser download wait timed out; no pass claim is made for those paths.
  • Change: added standard-library Node tests, extracted the active score-display and HTML-export seams to src/lab/, and added scripts/fixture-service.mjs. The fixture service records and verifies an opaque run/test receipt in a local JSON file; it is development-only and not a deployed multi-instance fixture backend.
  • Validation: npm test passed (3 tests); npm run typecheck passed; npm run build passed (Vite 6.4.4, dist/assets/index-COA8O12O.js); git diff --check passed.
  • Deployment status: no commit, push, deployment, provider action, or production fixture activation was performed.

2026-10-15 — In-app audit follow-up

  • Browser validation at http://127.0.0.1:8080 opened the Audit suite after the module extraction. The visible suite summary reported 15 / 15 passed assertions.
  • The local Export Escaped HTML action displayed the Exported adblocker-test-lab-v4.1.html success toast. The automation download-event wait remains unavailable, so no downloaded-file artifact path is recorded.
  • Deployment status: no deployment was performed.

2026-10-16 — Technical SEO foundations

  • Added a truthful canonical root URL, descriptive title and description, crawl directive, Open Graph/Twitter card metadata, and WebApplication structured data to the Vite document shell.
  • Added public robots.txt and sitemap.xml for the single public root route; documented post-deployment Search Console submission and the ranking/indexing boundary.
  • No keyword stuffing, fake reviews, provider changes, tracking, or misleading structured data were added.
  • Validation and deployment status: pending final typecheck, build, diff check, and local browser validation; no deployment performed.

2026-10-16 — Technical SEO validation follow-up

  • npm run typecheck and npm run build passed; git diff --check completed without diff errors. The production build contains the canonical link, Open Graph/Twitter metadata, WebApplication JSON-LD, robots.txt, and sitemap.xml.
  • Browser validation at http://127.0.0.1:8080 confirmed the updated title, description, canonical URL, robots directive, JSON-LD, and rendered Overview copy.
  • Deployment status: no commit, push, sitemap submission, Search Console action, or deployment was performed.

2026-10-16 - ExoClick ownership verification baseline

  • Added exactly one ExoClick ownership-verification meta tag to the document head. It is static verification metadata only: no ExoClick runtime loader, zone, provider request, or benchmark scoring behavior was added.
  • Added a built-document test that requires one tag with the expected ownership value; the test initially failed against the prior build (zero matching tags), then passed after the change.
  • Validation: npm run typecheck, npm run build, focused document-head test, full npm test (4 passing tests), and git diff --check passed. Local Chromium at http://127.0.0.1:8080 rendered the Overview; its document title remained correct.
  • Dashboard acceptance and production availability remain unverified. No provider request, creative, no-fill result, or filtering conclusion is implied by this metadata.

2026-10-16 - Real Ads Stress Lab configuration boundary

  • Added a pure configuration catalog for six optional Adsterra banner sizes, optional Adsterra Native/Popunder/Social Bar/Smartlink formats, and optional Monetag and ExoClick format lists. It accepts only valid HTTPS values and leaves malformed or absent values unavailable.
  • Added blank VITE_REAL_ADS_* variable names to .env.example and deployment guidance. No provider tag, zone ID, publisher identifier, Smartlink, or live script was committed.
  • Validation: focused configuration tests cover blank configuration, an individual banner, six distinct banners, and unavailable malformed Monetag/ExoClick input. Live-provider delivery remains unverified and no app UI starts a provider session yet.

2026-10-16 - Real Ads Stress Lab session ownership

  • Added an explicit client-side session lifecycle with IDLE, STARTING, ACTIVE, STOPPING, STOPPED, and ERROR states. It creates only app-owned mount/script nodes after a caller invokes Start, refuses duplicate Starts, and removes owned nodes/listeners/observers on Stop.
  • Popunder Stop responses flag that an already-triggered third-party effect can require a document reload; the implementation never clicks, reloads, or generates impressions automatically.
  • Validation: focused lifecycle tests cover idle/start/duplicate-start, cleanup and popunder reload flag, and replacement reload. No provider configuration was supplied and no live script was loaded during this validation.

2026-10-16 - Real Ads Stress Lab provider adapters

  • Added inert adapters for Adsterra, Monetag, and ExoClick. Adsterra exposes level-based banner/native/social/popunder composition, labels Smartlinks as manual-only, and excludes Popunder from compatible maximum mode. Monetag and ExoClick return no placements until valid configured formats exist.
  • Added sequential Adsterra option consumption so one supplied options object is set immediately before its corresponding placement start and removed afterward. The adapter never appends a script itself.
  • Validation: focused adapter tests cover six distinct banners, popunder exclusion, manual Smartlink handling, empty provider fallback, and sequential options. No live provider request or creative was observed.

2026-10-16 - Real Ads Stress Lab evidence normalization

  • Added independent statuses for not-started, script-loading, script-loaded, rendered creative, no-fill, request failure, possible filtering, provider error, and inconclusive observations. A rendered-creative label now requires a direct visible-layout observation.
  • Every normalized live-ad observation explicitly remains ineligible for the authoritative benchmark score and privacy-signal index.
  • Validation: focused evidence tests cover every status and reject hidden/indirect rendered claims. No provider creative was observed.

2026-10-16 - Real Ads Stress Lab local UI and Adsterra configuration

  • Change: added an explicit Real Ads Stress Lab settings view with Level 0-5, provider selection, Start/Stop/Clear controls, planned-placement disclosure, and independent placement evidence. The default is Level 0 with no provider initialization.
  • Provider and formats: local ignored configuration now contains the owner-supplied Adsterra six-banner inventory (160x300, 160x600, 300x250, 320x50, 468x60, 728x90), Native Banner, Popunder, Social Bar, and manual Smartlink. ExoClick has verification metadata only; Monetag has no approved live tag configured.
  • Configuration status: provider values are confined to ignored .env.local; no provider tag, zone identifier, Smartlink, or production value was added to tracked files. Banner loading is sequential around the shared atOptions value; Popunder remains Level-4 opt-in and Smartlink remains manual.
  • Verification performed: npm test passed (24 tests), npm run typecheck passed, npm run build passed, and local Chromium rendered the Lab control surface with zero configured formats before the development server reloads its ignored environment file.
  • Observed results: no live session was started by automation. Therefore no provider request, impression, no-fill result, creative, uBlock Lite comparison, or provider-attributed failure is recorded.
  • Errors and limitations: actual delivery requires a Vite development-server restart or an explicitly configured deployment environment, then a user-initiated session on the approved host. A loaded script remains insufficient evidence of a rendered creative.
  • Deployment status: no deployment, account change, dashboard action, commit, or push was performed for this local configuration work.

2026-10-16 - Real Ads Stress Lab configuration refresh observation

  • Verification performed: after the ignored local configuration was written, the existing Vite server at http://127.0.0.1:8080 rendered the Real Ads Stress Lab with 10 available configured formats while Level 0 remained selected and Start Session remained disabled.
  • Observed results: this confirms configuration discovery only. No provider level was started, no script was intentionally requested, and no creative/no-fill/blocking conclusion is recorded.
  • Errors and limitations: an attempted automation-only select action used an unsupported selector syntax and did not alter the page state. The user-owned Vite process was not stopped or restarted.

2026-10-16 - Real Ads Stress Lab comparison controls

  • Change: added explicit individual-format and comparison-viewport controls to the Real Ads Stress Lab. A selected viewport is evidence metadata only; it does not alter the visitor's browser dimensions. Placement evidence now preserves timestamp, format, dimensions, and viewport while remaining benchmark-ineligible.
  • Verification performed: added a focused evidence-context test; npm test passed with 25 tests, npm run typecheck passed, npm run build passed, and git diff --check completed without diff errors.
  • Observed results: no provider session was started during validation. No provider request, creative, no-fill, failure attribution, or uBlock Lite comparison is recorded.
  • Deployment status: no deployment, commit, push, dashboard action, click, reload, or impression generation was performed.

2026-10-16 - User-observed Adsterra live-delivery evidence

  • Observed result: user-provided Chromium evidence at local localhost:8080 shows the Real Ads Stress Lab reporting SCRIPT_LOADING and SCRIPT_LOADED for supplied Adsterra banner, Native Banner, and Social Bar placements, with multiple visible third-party ad creatives in the page and an on-screen Social Bar creative.
  • Attribution boundary: this is evidence that Adsterra delivery occurred in that local browser state; it does not establish a production result, a fill guarantee, or any conclusion about uBlock Lite because the screenshot is an unprotected-state observation only.
  • Provider status: ExoClick still has ownership-verification metadata but no supplied zone snippet. Monetag still has no configured approved live tag. Their absence from the shown session is a configuration gap, not evidence that either provider was filtered or returned no-fill.

2026-10-16 - Real Ads Stress Lab primary navigation, placement stage, and isolated loader

  • Primary navigation: promoted Real Ads Stress Lab (real-ads) directly into the application's primary top-level ribbon navigation alongside the Benchmark, Filter Analyzer, and Settings tabs, with its own dedicated route and icon (Zap). Removed it from nested settings sub-tabs to ensure exactly one authoritative route.
  • Placement stage: created an explicit, visible, dimension-locked Ad Placement Stage within RealAdsStressLab.jsx. All six configured Adsterra banner dimensions (728x90 Leaderboard, 468x60 Banner, 300x250 Medium Rectangle, 160x600 Wide Skyscraper, 160x300 Half Skyscraper, 320x50 Mobile Leaderboard), a dedicated Native Banner multi-card container, and status cards for Social Bar, Popunder, and manual Smartlink have reserved layout targets with real-time status badges.
  • Provider loader correction: updated session.js to support mounting synchronous Adsterra banner snippets inside isolated provider iframes within the designated target container. Each iframe encapsulates atOptions and script execution, preventing global scope collisions between banners, preventing parser-stream disruption, and ensuring clean teardown without document pollution or DOM leaks.
  • Sandbox containment: moved the hidden test fixture sandbox in AdBlockerTestLab.jsx out of the top navigation bar into a strictly-contained, off-screen root container, preventing fixture flashes and DOM leakage into unrelated app sections.
  • Evidence classification: enhanced src/lab/real-ads/evidence.js with inspectPlacementEvidence to evaluate real rendered creative dimensions and frames rather than passing unconditional false visibility. Accurately distinguishes SCRIPT_LOADING, SCRIPT_LOADED, REAL_CREATIVE_RENDERED (requiring verified positive layout dimensions), EMPTY_OR_NO_FILL, REQUEST_FAILED, BLOCKED_OR_FILTERED_POSSIBLE, PROVIDER_ERROR, and INCONCLUSIVE. All live-ad evidence remains strictly outside the authoritative 55-vector benchmark score.
  • Teardown & session safety: implemented unmount cleanup in useEffect and duplicate-start prevention via active refs. Stop Session disposes of all provider iframes and scripts, resets slot statuses, and alerts if an active Popunder requires a document reload.
  • Tests and verification executed:
    • npm test executed and passed (29 unit tests passing across all test suites, including new tests for isolated provider iframe mounting and evidence inspection).
    • npm run typecheck executed and passed with 0 errors.
    • npm run build executed and passed (Vite production bundle built cleanly).
    • git diff --check passed with 0 whitespace issues.
  • Observed local behavior: navigation ribbon cleanly switches to Real Ads Stress Lab; placement stage renders organized, dimension-locked slots awaiting session start; no automated network requests or impressions occur until the user explicitly clicks Start Session.
  • Provider limitations: Adsterra remains the only provider with local live-delivery configuration; ExoClick has verification metadata only, and Monetag has no approved live tag configured.
  • Deployment status: no deployment, commit, push, dashboard modification, or live credential creation was performed.

2026-10-16 - UI/UX Overhaul execution (Phases 4 through 8)

  • Phase 4 (Overview & primary run flow): simplified the hero section with an outcome-first headline ("Inspect browser behavior without inventing the cause"), prominent Quick Scan (6 checks, ~1.5s) primary CTA, Full Diagnostic (55 checks, ~8s) secondary CTA, and a 3-step visitor guide ("Controlled Fixtures", "Objective Telemetry", "Unbiased Audit Score"). Implemented a true first-run empty state explaining score standby instead of an artificial 0/100. Moved browser context diagnostics into an expandable <details> disclosure.
  • Phase 5 (Results & workbench comprehension): restructured Results & Workbench to present coverage and denominator before the headline score. Added a dynamic plain-English insight observation banner, text-backed status indicators (BLOCKED, PARTIAL, ALLOWED, INCONCLUSIVE), visible "Reset Filters" pill, and toast notifications for JSON/CSV/HTML exports.
  • Phase 6 (Accessibility & responsive quality): verified keyboard navigation with visible focus rings (focus-visible:ring-2 focus-visible:ring-[#c65322]), added role="status" and aria-live="polite" on active execution HUD, wrapped data tables in horizontal scroll containers, and verified responsive layout reflow across mobile (320px–390px) through desktop breakpoints without horizontal body overflow.
  • Phase 7 (Controlled architectural modularization): safely extracted UI components from AdBlockerTestLab.jsx into modular components under src/components/:
    • src/components/NavigationRibbon.jsx: primary navigation header, score chips, run controls, theme toggle.
    • src/components/OverviewView.jsx: hero, scan cards, first-run state, and collapsible browser context.
    • src/components/ResultsWorkbenchView.jsx: HUD, score cards, category breakdown, export toolbar, filter bar, results list, and 6-stage audit drawer.
    • src/lab/navigation.js: pure route normalization and 5 primary ribbon destination constants.
    • Added unit test suite test/navigation.test.mjs verifying destination counts and alias normalization.
  • Phase 8 (Verification & quality gate):
    • npm test executed and passed: 31/31 unit tests passing across all suites.
    • npm run typecheck executed and passed: 0 TypeScript errors.
    • npm run build executed and passed: production bundle compiled cleanly in 5.32s.
    • git diff --check executed and passed: 0 whitespace issues.
  • Attribution and provider limitations: no live provider credentials or secret tags committed; no automated ad clicks, impressions, or popunders generated; live-ad telemetry remains strictly non-scoring.
  • Deployment status: no production deployment, commit, push, or credential modification was performed.