Surfaced during the whole-branch review of #618 (PR 2). Pre-existing, deliberately left out of scope there.
Problem
crates/rdlp-postprocess/src/pipeline/stages/recode.rs:490:
opts.video_codec.as_deref().unwrap_or("libx264")
On the remux fast path opts.video_codec is always None — remux copies the video stream rather than choosing an encoder. So the .unwrap_or fallback always fires, and every remux stamps the literal "libx264" into the encoding_tool metadata regardless of what the input actually contains.
Remuxing a VP9 file to MKV writes encoding_tool: "rdlp/<version> (libx264 + ...)" — claiming an encoder that never ran, for a codec that isn't there.
Why it wasn't fixed in #618
PR 2 fixed the audio half of the same tag (a video-only recode was stamping "+ copy" for audio it did not have, and a remux was doing the same). The video half is the same class of untruth but a different value and a different data path, so it was filed rather than folded in.
Fix shape
Thread the probed media_info.video_codec into the remux encoding_tool capture instead of the hardcoded fallback. The probe already runs on this path and the value is in hand.
Acceptance
Refs #618
Surfaced during the whole-branch review of #618 (PR 2). Pre-existing, deliberately left out of scope there.
Problem
crates/rdlp-postprocess/src/pipeline/stages/recode.rs:490:On the remux fast path
opts.video_codecis alwaysNone— remux copies the video stream rather than choosing an encoder. So the.unwrap_orfallback always fires, and every remux stamps the literal"libx264"into theencoding_toolmetadata regardless of what the input actually contains.Remuxing a VP9 file to MKV writes
encoding_tool: "rdlp/<version> (libx264 + ...)"— claiming an encoder that never ran, for a codec that isn't there.Why it wasn't fixed in #618
PR 2 fixed the audio half of the same tag (a video-only recode was stamping
"+ copy"for audio it did not have, and a remux was doing the same). The video half is the same class of untruth but a different value and a different data path, so it was filed rather than folded in.Fix shape
Thread the probed
media_info.video_codecinto the remuxencoding_toolcapture instead of the hardcoded fallback. The probe already runs on this path and the value is in hand.Acceptance
encoding_tool, notlibx264.Refs #618