Analyze visitor screen-size/device cohort (Vercel Web Analytics) → optimize Pluto layout/breakpoints to dominant cohort, VERIFIED vs actual distribution (telemetry directive bs-mqmscr4k3wh)

Sub-tickets

Questions

Activity

• wi-cli-venus created · 2026-06-20
• wi-cli-venus blocked · 2026-06-20
  Blocked on Elazar A/B (raised in bs-mqmscr4k3wh seat 3): A=analyze Vercel Web Analytics coarse device cohort (no new collector — Pluto has no custom screen-size pipe); B=build /api/log-client-env to the converged 4-field allowlist (viewportBand,deviceType,osFamily,orientation) for true viewport bands first. WI scope (data source + whether a build precedes) depends on his pick. Also note Pluto minimization candidate raised: drop ip+userAgent on non-security info events (toast/404/navigation), keep on auth/authz/malicious.
• wi-cli-venus commented · 2026-06-20
  A/B reframed by pm-venus (seat 4): option B is CHEAP — Pluto clones venus's already-built+minimized banded pipe (viewport-reporter → telemetry-action → appEvents detail jsonb → device-mix reader, k-anon k=5), NOT greenfield. Lowers B's cost substantially; pm-venus recommends B-cheap (Vercel-coarse A exports poorly + breaks fleet uniformity). CONVERGED ALLOWLIST (fleet): {widthBand, deviceType, osFamily, orientation} core + dprBucket per-app consumer-gated + role[] banded (replaces actorUserId — no PII). Read-side k-anon k=5 on any snapshot/export. Still blocked on Elazar A/B; if B, scope = port venus pipe + apply k-anon, then later stop-condition.
• wi-cli-venus commented · 2026-06-20
  FIELD-NAME LOCK (addendum to bs-mqmscr4k3wh): canonical width field = viewportBand (NOT widthBand) fleet-wide. Pluto adopts viewportBand under option B. A/B = Elazar-pending, room silence-default B-cheap (generalpm rolling to Elazar; my A/B rec DM also sent).
• wi-cli-venus commented · 2026-06-20
  RE-SCOPED by brainstorm bs-mqmt93jl25b (Elazar, 2026-06-20): Pluto builds NO screen-size collector. A/B (Vercel-coarse vs B-cheap venus-clone) CLOSED/MOOT. New scope: optimize Pluto's ALUMNO-facing routes (/mi-cursada, student views) to MARS's robust envelope — mobile-first, portrait-primary, 320-447px phone target; design to the ENVELOPE, NOT Mars's 384-415px pixel peak (Argentina-undergrad/iOS-specific, over-fit on 5-day/709-session window). Pluto-specific boundary I added + room accepted: DOCENTE/ADMIN surfaces (/informes dense tables, /alumnos, /admin) are desktop/tablet-leaning by usage and stay desktop-first — do NOT apply the alumno mobile envelope to grading dashboards. Same-cohort validity: Pluto = same UBA odontología population as Mars. Any future Pluto-own collection = optional non-blocking verification. Pending pm-venus turn + opener kind=decision close; will reflect final on close.
• wi-cli-venus commented · 2026-06-20
  FINAL (bs-mqmt93jl25b CLOSED 4/4): Adopt Mars envelope on Pluto ALUMNO routes (mi-cursada/student) — mobile-first, portrait-primary, 320-447px phone envelope; design to envelope, NOT the 384-415px peak. DOCENTE/ADMIN (informes/alumnos/admin) stay DESKTOP-first (Mars=alumno signal, wrong prior for grading dashboards). NO collector built. Own collection optional/non-blocking. Ready for coder dispatch (no longer blocked on A/B — that's closed/moot).
• wi-cli-venus statusChanged · 4w ago
  Re-scoped per bs-mqmt93jl25b close: no A/B collector, optimize alumno routes to Mars mobile envelope 320-447px portrait, docente/admin desktop-first. Dispatch-ready, not gate-blocked - capacity-parked behind coder-pluto-cc's current queue.
• wi-cli-venus completed · 1w ago
  Superseded by PLUTO-537 (done 2026-07-10), same telemetry directive bs-mqmscr4k3wh. Two independent reasons, either sufficient: (1) the named source does not exist — Vercel Web Analytics is NOT enabled on prj_0t8ww1LnJr4IzXrKzLFr56b4mI5t (get_web_analytics -> 404 'Web Analytics not found'), and pluto-commons records Speed Insights as deliberately uninstalled, so there is no Vercel-side cohort data to analyze and never was. (2) The question was already answered from a BETTER source: PLUTO-163's banded viewport telemetry in appEvents (auth-gated real users, role-tagged, one row per login — not crawler-polluted pageviews). PLUTO-537 pulled it: 427 rows / 295 users, 87% mobile, dominant band 384-415px portrait = 71% of mobile, alumnos and docentes both mobile-heavy; desktop ~1344-1535 landscape. Step 2 acted on it (3 tap-target fixes, 2fdfd05 v2.5.5). Collection was then stopped deliberately (PLUTO-536, Elazar 2026-07-07) because the campaign was complete. Nothing left here that PLUTO-537 did not deliver. Residual layout work lives in PLUTO-40 (base-up mobile-first refactor), which is a different scope from a cohort measurement.