Releases
v2.107.48
Compare
Sorry, something went wrong.
No results found
Chore : Retain browser append and media-fragment parse samples for the playback forecast.
Chore : Keep first-required-segment completion unavailable until browser append duration has a process-based model.
Chore : Record browser load-to-parse and append-request-to-acknowledgement as separate unknown predecessors in first-segment diagnostics.
Fix : Keep first-segment readiness unknown when its media ranges are missing instead of treating an empty range list as buffered.
Chore : Include per-fragment parsing and buffering spans in the playback model records.
Chore : Join recent browser work samples to their generated output segment by HLS sequence number.
Chore : Write a bounded playback forecast record when browser work samples change, not only when readiness changes.
Chore : Define the completion point under absolute relative error without fitting a completion-time distribution.
Chore : Include request-received, readiness-snapshot and response-prepared proxy-clock marks in progress responses.
Chore : Include worker-read and response timing marks with source statistics so the browser can bound their age.
You can’t perform that action at this time.