test(qa): sprite and enrichment checks on both engines through the probe

The QA probe gains drawn(layerId): what a layer draws per object (glyph
image, scale, shown, screen rotation, alpha, position; its 3D model's
scale, shown, ready), from the flights layer's qaDrawn(), which reads the
renderer's handles, so it is the same on every engine.

qa-sprites-b5 and qa-enrich-ambient take --engine=cesium|navara: without a
viewer they read the layer through the probe and move the engine camera;
the military rows/cluster are skipped where the engine has no military
layer. Cesium verdicts unchanged (sprites 9/9; enrich 9/2, the same E8/E10
failures as on 7c7422e2); Navara: sprites 7/7 (flights rows and 3D cluster),
enrich 9/2 (the same two). Task 10b, checklist item 7.
This commit is contained in:
German Khokhlov committed 2026-10-17 03:15:12 +04:00
1 parent 03a0ff3a01
commit 132a941e9a
5 files changed
+135 -46

No files matched your search

+3
View File
@@ -11,6 +11,8 @@
// plain { x, y, z }, screen a window point or null)
// screenOf(id, objectId) the window point of one object, or null
// selected(id) the layer's tracked/selected info, or null
// drawn(id) what the layer draws per object (glyph, model),
// as plain data, or null (layers with qaDrawn)
// following() { label, native } of engine.camera.following()
// movers() engine.movers.list()
// framesDrawn() frames drawn since the probe started
@@ -67,6 +69,7 @@ export function createQaProbe({ engine, dataManager, availability }) {
return row ? screenOf(row.position) : null;
},
selected: (layerId) => moduleOf(layerId)?.getTrackedInfo?.() ?? null,
drawn: (layerId) => moduleOf(layerId)?.qaDrawn?.() ?? null,
following() {
const handle = engine.camera.following();
if (!handle) return null;