-
Notifications
You must be signed in to change notification settings - Fork 12
Video rebuffers after each ad #2
Comments
We have no other choice but to reuse player if we want this to work on iPhones. |
We could use different players by default and then fall back to reusing the single player on devices that require it. Discarding downloaded video data is not good default behavior. |
That would add a lot of complexity, for very little gain. My testing showed that iPhone did not work at all, iPad required manually clicking play whenever you switched player, and I also seem to remember that in fullscreen, you had the same situation as on iPhone. We have not done extensive Android, as Android is currently outside of the scope. |
I would argue there is quite substantial amount of gain. Forcing rebuffering of a video after every ad is a problem - especially with mobile connections becoming more common even on laptops. Also, Android is outside of scope? |
It's not a substantial gain, if no devices will benefit from it. Android is outside of scope currently, but will be supported in the not too distant future. Numbers rule. |
Current status: Preroll and postroll works great on all devices. Midroll is not optimal for long videos on mobile devices. |
With commit 1506b5f now reusing the original video player and changing the source will interrupt buffering, causing the video to start buffering from the seek time when it is resumes playback after the ad even if the user had loaded the entire video before starting it.
Users will probably be quite annoyed if they buffer up a long-ish video (which are usually the ones with ads) to watch (especially on a slow connection like, say, mobile) and once they hit the first ad, the video will need to load again... Also makes for quite a lot of wasted bandwidth.
In fact, iOS and Android are precisely the cases where we'd want to avoid this behavior since desktop OSes usually have a "proper" network connection and won't mind rebuffering.
The text was updated successfully, but these errors were encountered: