/fly Manual Test Checklist

A walk-through of representative missions to exercise every code path in the heliocentric + cislunar /fly scenes. Use this as the manual smoke test after any /fly change, and as the reference spec for the future automated e2e equivalent.

The checklist was first executed on 2026-06-09 after the cinematic camera + FD stage marker rework (commit 7a8d8e* → main). Results + known-broken items at the bottom.

---

How to drive it

1. npm run dev (vite preview on localhost:5273).
2. Open Chrome DevTools MCP or just use a regular browser tab.
3. For each mission below, open the URL, scrub the timeline slider to the listed checkpoints, and verify the expectations.
4. Open the browser console — every test below expects zero error or warn messages.

Useful one-liners (paste into DevTools console):

• Count revealed FD chips: document.querySelectorAll('[data-testid="fd-phase-marker"]').length
• List revealed chip labels in order: Array.from(document.querySelectorAll('[data-fd-phase]')).map(e=>e.dataset.fdPhase)
• Jump the scrubber to a given progress (0..1):

  const s = document.querySelector('input[type="range"]');
  const set = Object.getOwnPropertyDescriptor(HTMLInputElement.prototype, 'value').set;
  set.call(s, '0.75'); s.dispatchEvent(new Event('input', { bubbles: true }));

---

Test cases

A. Round-trip heliocentric — ORRERY-1 demo

URL: /fly (no params). Trajectory: Heliocentric free-return Mars flyby. arr_day = 509.

Expectations:

• 3D scene renders Sun at centre with both Earth + Mars orbits visible.
• Three anchor rings:
  • LAUNCH (blue, "Earth dep") at outPts[0].
  • ARRIVAL (red, "Mars arrival") at outPts[last].
  • RETURN (blue, "Earth arrival") at retPts[last].
• FD chips on the arc, revealing on schedule:

  scrubber	revealed	notes
  0.0	Injection	tiny window before CRUISE threshold (0.03)
  0.05	Injection, Cruise	CRUISE just past INJECTION
  0.45	Injection, Cruise	mid-outbound
  0.50	Injection, Cruise, Approach	outbound APPROACH fires
  0.55	Injection, Cruise, Approach, Arrival	Mars-ARRIVAL fires
  0.70	+ Cruise (return)	return CRUISE fires after Mars flyby
  0.92	+ Approach (Earth)	return APPROACH fires
  0.99	+ Arrival (Earth)	all 7 stages revealed

• Camera arc:
  • t=0: tight on Earth (~30u).
  • Slow LERP into wide cruise; Sun visible, ship tracked.
  • Approach-Mars (~outboundT 0.8): closeup on Mars destination.
  • Mars-depart (returnT < 0.05): held shot tracking ship as it leaves Mars.
  • Earth-approach (~returnT 0.9): closeup on Earth.
• FlightDirectorBanner phase eyebrow cycles through INJECTION → CRUISE → APPROACH → ARRIVAL → CRUISE → APPROACH → ARRIVAL.
• No console errors.

B. One-way Mars surface — curiosity

URL: /fly?mission=curiosity. Trajectory: Heliocentric Hohmann. dep=2011-11-26, arr=2012-08-06.

Expectations:

• LAUNCH ring (blue) at Earth-at-dep; ARRIVAL ring at Mars-at-arr.
• ARRIVAL ring subtitle uses mission.arr_label directly ("2012-08-06"), NOT the destination-aware override (that only triggers on round-trip).
• No RETURN ring (retPts.length === 0).
• FD chips:

  scrubber	revealed
  0.05	Injection, Cruise
  0.5	Injection, Cruise
  0.85	+ Approach
  0.97	+ Arrival

• FD banner cycles only INJECTION → CRUISE → APPROACH → ARRIVAL (never enters the return-leg branch).
• No console errors.

C. Round-trip asteroid sample-return — hayabusa2

URL: /fly?mission=hayabusa2. Trajectory: H-IIA → Ryugu rendezvous + sample return.

Expectations:

• LAUNCH (blue) at Earth-at-dep, ARRIVAL ring positioned at the destination (asteroid), with subtitle Asteroid arrival (the destination-aware override for round-trip).
• RETURN ring (blue) at retPts[last] with subtitle Earth arrival.
• Full 7-stage FD chain reveals across timeline.
• Approach-Earth on return uses HELIO_EARTH_CLOSEUP_R=50 (looser Earth zoom than destination-side HELIO_CLOSEUP_R=40).
• No console errors.

D. Outer-system one-way (Jupiter / Saturn range) — voyager-1

URL: /fly?mission=voyager-1. Destination: Saturn (9.5 AU).

Expectations:

• 3D scene renders Sun + Earth orbit + Saturn orbit, all visible at the wider cameraDistanceFor(SATURN) = max(180, 9.5·80·2) = 1520u.
• 4 outbound FD chips reveal as scrubber moves through 0 → 1.
• No RETURN ring (one-way).
• Long cruise gives the steady-cam azimuthal drift (0.05 rad/s) time to read — camera should slowly orbit the spacecraft.
• No console errors.

E1. One-way Moon — slim

URL: /fly?mission=slim.

Expectations (cislunar one-way):

• Cislunar scene active.
• helioFd === 0 (FD diamonds gated out).
• cislunarMarkers === 1.
• ARRIVAL event at MET 110 — no EARTH RETURN. retPts.length === 0.
• No console errors.

E. Moon cislunar — apollo11

URL: /fly?mission=apollo11.

Expectations:

• Cislunar scene active (Earth-centred, Moon orbits at exaggerated scale).
• Helio FD diamonds do NOT render (gated by viewMode === 'heliocentric').
  • DOM check: document.querySelectorAll('[data-testid="fd-phase-marker"]').length === 0.
• Cislunar phase markers DO render (separate pipeline).
  • DOM check: document.querySelectorAll('[data-testid="phase-markers-overlay"]').length === 1.
• Apollo timeline events visible in CAPCOM panel (LAUNCH / TLI / LOI / TEI / EARTH RETURN).
• No console errors.

F1. Modern lunar round-trip — artemis3

URL: /fly?mission=artemis3.

Expectations:

• Cislunar scene active (same path as Apollo 11).
• helioFd === 0, cislunarMarkers === 1.
• Events span LAUNCH → TLI → TCM → ASCENT → TEI → EARTH RETURN — full SLS/Starship crew-and-return cadence.
• No console errors.

F. Cislunar free-return — apollo13

URL: /fly?mission=apollo13.

Expectations:

• Cislunar scene active. Same constraints as Apollo 11 (no helio FD, yes cislunar phase markers).
• Free-return arc visible.
• CAPCOM panel surfaces the anomaly event.
• No console errors.

---

Bonus checks (do these manually on at least one round-trip mission)

• Toggle Science Lens on. The top-centre Science Lens panel must sit directly under the nav header (NOT in the canvas centre — that regression was caused by FlightDirectorBanner publishing its height to --lens-banner-height while sitting in the bottom strip).
• Toggle the timeline Pause / Play button. Camera lerps should pause / resume cleanly.
• Drag the canvas to orbit the camera; release. Steady-cam drift should stop on drag and resume after.
• Pinch-zoom (or wheel-zoom). User-zoom should override the auto-zoom for the rest of the sub-phase.
• Reload with prefers-reduced-motion: reduce. Camera drift + fd-marker pop animation should disable.

---

G. Flyby Cinema — voyager-2 (and other grand-tour missions)

URL: /fly?mission=voyager-2. Mission events include 4 flyby entries: Jupiter (MET 688), Saturn (MET 1466), Uranus (MET 3079), Neptune (MET 4388).

Expectations (per flyby):

• When |simDay − dep_day − met_days| ≤ 25, the camera enters a dedicated flyby-<met> sub-phase.
• Sub-phase: closeup framing (HELIO_FLYBY_R = 80u) centred on the live spacecraft scene position with approach-pitch tilt.
• Sun visible in frame (since spacecraft is at heliocentric position of the flyby planet, ~5–30 AU from origin).
• Smooth LERP in / out of the closeup as the window opens + closes.
• Outside the window, regular cruise / approach sub-phases run.

How to drive:

// Jupiter flyby at MET 688 of 4388 total → slider ≈ 688/4388 = 0.157
const s = document.querySelector('input[type="range"]');
const set = Object.getOwnPropertyDescriptor(HTMLInputElement.prototype, 'value').set;
set.call(s, '0.157'); s.dispatchEvent(new Event('input', { bubbles: true }));
// Pause via the Pause button, then wait a few seconds for camera lerp.

Same pattern is data-driven: it auto-activates for any mission whose flight.events[] includes one or more type: 'flyby' entries. Tally from a 2026-06-09 grep across mission JSON files:

Mission	Destination	Flybys	Cinema value
bepicolombo	Mercury	9	🏆 highest
messenger	Mercury	6	high
rosetta	Comet	6	high
galileo	Jupiter	5	high
cassini	Saturn	4	high (✓ verified)
voyager-2	Neptune	4	high (✓ verified)
juice	Jupiter	4	high
new-horizons	Pluto	3	mid
giotto	Comet	3	mid
voyager-1	Saturn	2	mid
pioneer-11	Saturn	2	mid
vega-1 / vega-2	Venus	2	mid
pioneer-10, mariner4, ulysses, hayabusa2, dawn, juno	various	1	low (single arrival)

Cassini Earth-flyby validation: scrubbing to MET 894 (slider ≈ 0.365) on a paused sim, then clicking the timeline "FLYBY at MET 894.00 days" jump button, drives the camera into a tight closeup that shows the spacecraft and the LAUNCH ring at Earth's position simultaneously.

---

H. Milestone overlay — labeled flight.events[]

URL: /fly?mission=cassini (the first backfilled mission).

Expectations:

• Each flight.events[] entry that carries a label field renders as a small teal ◇ <label> chip at the spacecraft's projected scene position at the event MET.
• Distinct visual language from FD stage markers: teal vs gold, chip-with-leading-diamond vs chip-above-leader, no reveal gating (milestones always render once the event is on-screen).
• DOM contract:
  • Overlay container: [data-testid="milestone-overlay"] with data-milestone-count="<N>".
  • Each chip: [data-testid="milestone-chip"] with data-met-days="<met>", contains .milestone-tick (◇) + .milestone-label (text).

Driving Cassini through its milestone roster:

• Jump to MET 0 → "◇ Launch" near LAUNCH ring.
• Jump to MET 207 → "◇ Venus #1 — gravity assist" at the spacecraft.
• Jump to MET 808 / 894 / 2226 → Venus #2 / Earth / Jupiter chips.
• Jump to MET 2451 → "◇ Saturn orbit insertion" at Saturn.

Iconic-mission backfill status (2026-06-09):

• ✅ Cassini (6 milestones)
• ⏳ All 17 other iconic missions still anonymous — backfill is a mechanical pass joining /static/data/trajectories/<id>.json waypoints to /static/data/missions/<dest>/<id>.json events by date. Tracked as a follow-up data migration.

---

Round-2 changes (2026-06-09)

These are the broken items from the initial run that have been fixed or addressed in this same session:

1. Helio camera far-plane was 4000u, clipping Neptune + Pluto. FIXED.

   • Bumped PerspectiveCamera.far from 4000 → 16000 in src/lib/three/fly-helio-scene.ts.
   • Also capped cameraDistanceFor at 1600u (Saturn-scale wide) in src/lib/fly-scene-constants.ts — Voyager 2 / Pluto missions keep the Saturn-level framing and the spacecraft simply flies off the wide-frame edge during cruise, approached only when the flyby cinema sub-phase zooms in on it.
   • Outer-system visual readability is still poor (Sun + orbit rings read as small at 1600u framing) — that's a separate LOD-scaling UX pass beyond this fix.

2. Hayabusa 2 mission data clipped to primary Earth-return. FIXED.

   • Removed the extended-cruise events at MET 4250 + 6092 from static/data/missions/asteroid/hayabusa2.json.
   • The visible mission events now stop at MET 1880 (sample-return capsule reentry over Woomera).
   • Note: a full mission-length clip would require an end_met_days architectural addition (transit_days × 2 is still the round-trip total length internally). Filed as a follow-up.

3. One-way Moon-mode coverage added. See test E1 (SLIM) above.

4. Flyby Cinema mode added for grand-tour missions. See test G.

---

Still to-fix-next (carry forward)

1. Outer-system visual LOD. With cameraDistanceFor capped at 1600u and the camera framing wide, the Sun (8u radius) projects to a few pixels and the orbit rings disappear into the background. A distance-aware scale-up (or LOD swap) for the Sun mesh + orbit line widths would make outer-system framings actually readable.

2. end_met_days override field on missions. Hayabusa 2 needs to end at MET 1880 (Earth return), but the scrubber still spans the round-trip-derived 2604 days. Adding an explicit flight.end_met_days field would let mission data clip the timeline without abusing transit_days.

3. Flyby planet meshes. The flyby cinema sub-phase zooms close on the spacecraft, but the flyby body (Jupiter, Saturn, etc.) isn't rendered as a textured sphere — only the active destination is. Rendering all bodies on the trajectory's flyby roster would complete the "mini-movie" feel by giving the camera a planet to look at during the flyby beat.

---

Suggested next iteration of this doc

• Convert the per-mission scrubber tables into a Playwright spec (tests/e2e/fly-stages.spec.ts) using the same chip-reveal DOM selectors. Each row in the table → one expect(...).toHaveText(...).
• Add a programmatic "no console errors" assertion via page.on('console').
• Render the test plan inline at the bottom of tests/e2e/_helpers/fly.ts so contributors find it without leaving the spec.

---

Last manually executed: 2026-06-09 — Marko + Claude (Opus 4.7 · 1M context).