Read tracking invariants through the follow service, draw-port sets and layer records

The tracking harness had drifted off the app before this migration began and
stopped at its share-link reload: it read Cesium's trackedEntity record, which
the camera port's follow now owns as a hidden helper. Read every invariant
through the mechanism that replaced the one it used:

- what the camera follows (share links, jitter, M3 release, stale cue, weld,
  display floor) through data/followed.js, whose target's position is the
  same function the tracked entity's position ran;
- fleet icons from the draw port's icon sets as the renderer holds them (the
  GPU rows' colour, alpha, scale, shown flag, depth distance, position and
  drawn course), the tracked icon's own set left out as before;
- models by their {set, id} pick ids; H1 now also requires the hit to be the
  flights tracked-model set;
- display-floor/loading plants shown-but-not-ready on the layer's own model
  record, which is what the layer decides ownership from;
- display-floor/hold reads the icon after the layer's microtask write to its
  draw set.

Arrival rotation no longer has a CPU pass to tamper with or repair: the GPU
icon sets turn each icon along its drawn course every frame. The group now
measures the on-screen nose the drawn course projects to against the feed
course's, on the first frame after moveEnd and on the first frame drawn after
a frame that drew the row hidden (same 10 degree bound and budgets). That the
shader turns an icon to its course's screen angle is the renderer's contract,
drawn by the conformance reference scenes.

No tolerance changed and no invariant was removed. The pre-migration build
and this branch both pass 109 checks; a run with the grounded scale and the
fleet rotation mode broken fails the ground and arrival checks.
This commit is contained in:
Sameh Khamis committed 2026-11-08 00:51:37 -08:00
1 parent 38e0750715
commit fe79eb840e
2 files changed
+300 -178

No files matched your search

+1 -1
View File
@@ -5750,7 +5750,7 @@ pins and share links already supply complete boxes.
- Fleet model eligibility is distance-based with on-screen priority and hard caps (`MODEL_MAX`, `MODEL_MAX_ALL`) to avoid draw-call explosions. Each model owns its own `modelMatrix`; shared scratch matrices are forbidden because they caused stacking/flicker.
- Tracked aircraft use standalone model primitives driven from the already-settled dead-reckoned display position, while the tracked Cesium entity remains billboard-backed so `viewer.trackedEntity` always has a ready bounding sphere.
- Flights and military layers mirror the same tracking invariants: no warm-up freeze/jump, altitude-scaled framing, trail head glued to the displayed plane, no pull-out when switching targets, and no cross-layer orphan when switching between commercial and military tracks.
- Regression surface: `npm run test:track` drives the real app headless with synthetic aircraft feeds and asserts the tracking invariants without depending on live OpenSky/adsb data. Arrival-rotation checks observe the settle frame and then the first reveal frame, each within a 500 ms wall-clock budget that sits inside the one-second idle fallback, rather than assuming the 80 ms fleet cadence fits inside four rendered frames; an incorrect first orientation fails immediately. Only an inconclusive timeout is retried (once); a timeout keeps every angle it measured (the settle frame before a reveal timeout, and the late frame itself, which is read before the deadline check), so a measured wrong angle fails without retry. `src/data/trackedModelRegime.test.mjs` pins the tracked contact's threshold math, the enter/exit asymmetry, and the default-on / cockpit / TR-3B / deselect wiring in both layers.
- Regression surface: `npm run test:track` drives the real app headless with synthetic aircraft feeds and asserts the tracking invariants without depending on live OpenSky/adsb data. The harness reads what the app draws through its own seams: what the camera follows through `src/data/followed.js`, icons from the draw port's icon sets as the renderer holds them, and models by their `{set, id}` pick ids. Arrival-rotation checks read the drawn rows' course (the icon sets turn each icon along it on the GPU every frame, so there is no CPU rotation pass to repair a stale nose) and observe the settle frame and then the first reveal frame, each within a 500 ms wall-clock budget; an incorrect first orientation fails immediately. Only an inconclusive timeout is retried (once); a timeout keeps every angle it measured (the settle frame before a reveal timeout, and the late frame itself, which is read before the deadline check), so a measured wrong angle fails without retry. `src/data/trackedModelRegime.test.mjs` pins the tracked contact's threshold math, the enter/exit asymmetry, and the default-on / cockpit / TR-3B / deselect wiring in both layers.
### TR-3B conversion Easter egg (August 2026)