crates/namir-clap/src/params_ext.rs:305-306 vs :178-190
apply_flush_events's doc comment claims it is "shared between the main-thread and audio-processor flush implementations below", but the audio-processor impl hand-rolls the identical decode loop rather than calling it. Two copies of the same event decode is exactly the drift the helper exists to prevent, and which audio.rs:88-91 correctly avoids for apply_direct_and_mirror.
Fix: call it, passing an apply closure over apply_direct_and_mirror.
crates/namir-clap/src/params_ext.rs:305-306vs:178-190apply_flush_events's doc comment claims it is "shared between the main-thread and audio-processorflushimplementations below", but the audio-processor impl hand-rolls the identical decode loop rather than calling it. Two copies of the same event decode is exactly the drift the helper exists to prevent, and whichaudio.rs:88-91correctly avoids forapply_direct_and_mirror.Fix: call it, passing an
applyclosure overapply_direct_and_mirror.