Luxar Viewer API Documentation - v2026.9.22
    Preparing search index...

    Module core/app/init/load-activity

    The adaptive-DPR controller's "data is loading" predicate.

    While the viewer is not settled, frame times measure decode/upload/commit work, not steady-state render cost, so the controller must not learn from them (no probes armed or settled, no refresh-cap samples, no floor evidence). The controller used to read only isUpdateInProgress() — the update-view lock — which excludes the post-load progressive refinement drain (each rung is a fetch + decode + commit with the lock RELEASED between passes), lazy lod_group level loads, held partition rising-edge resyncs on the default loader, and load passes on non-default loaders. The 2026-09 audit caught the controller crediting the end of a refinement drain to a DPR step it had just taken.

    This is the same settledness the perf probes read through __luxarDebug.getPerf().isSettled (debug/perf-snapshot.ts), inverted.

    LoadActivityDeps
    LoadActivitySources
    isLoadActivity
    refinementCompleteFromTimeline
    buildLoadActivityPredicate