Closed tonihei closed 4 years ago
adBreakStart does not imply that anything is playing, it only signals that there is a change from main content twards ad content. There might still be initial buffeirng going on (now for the ad) before something is actually shown. Thus the initialStartupTime definition in the doc should be correct, it measures from the moment the user wants to play something, until something starts to play on the screen.
After the initial startup, if there are other playback interruptions (even between successive ads), these are not treated as "initial", but counted as rebufferings. Note that rebufferings can happen in or between ads as well as in the main content.
Just for clarification:
No, InitialStartupTime does not include the duration of the preroll ad. IntialStartupTime is from when the user requests content to when preroll ad starts to play.
The drawings are accurately capturing what we are intending. However the WG agrees to provide additional text to clarify the calculations.
First comment has been addressed in CTA-2066.
@njadia Update the diagram for "Playback with 2 ads in succession" with:
Completed, please close.
Action: Neel to update the notes section on the diagram included in appendix regarding sub sequence startup times. - CLOSED
Note: InitialStartupTime metric only includes the duration marked as "initial startup". The subsequent "startup" duration in the diagram above can be used to create additional metrics beyond the fundamental set of metrics targeted for this version of the spec.
WG agreed to add the "note" to each appropriate diagram.
initialStartupTime on p18 is currently defined for cases where no preroll or midroll ad is played.
Based on the diagrams in the appendix, the actual definition consists of 2 parts:
and only in absence of these two: the time between playbackRequest and playbackStart.