thexeos
commented
6 years ago This can definitely be a use case for MediaSession API. Right now, it can be tricky to let the person know about the incoming call. If no sound has been played from the tab, and the tab is in the background (and on mobile, if no user gesture was present), there is no way to sound the ringtone. The only way to notify of the call is to use the Notifications API.
Problems with Notifications API approach are: 1) No way to play the sound together with the notification (and sound from the tab may be desynchronized with the notification) 2) Limited control over how the action buttons look. On some platform the buttons are stacked (one per row) and on others they are displayed two per row. 3) No direct contact with the page - the action button clicks are handled in the Service Worker, which then has to pass the data to the tab that requested the notification (or in case it was triggered by Web Push API, then the page has to be opened first, then the page has to request the data from the service worker, since there is no way that I know of, which would allow notificationclick event handler to reliably stay running while the page is being loaded, and then pass the newly loaded page the click data)
Having MediaSession API provide built-in functions would definitely help.
For the call initiation:
- Ability to play system default (or developer-provided) ringtone
- Reliably wake screen (if supported/enabled) when receiving the call
- Predictable UI and toggleable elements (developer won't be required to provide icons or name buttons, which might produce i18n problems too)
And during the call:
- Ability to quickly hang up, by tapping a "notification"
- Ability to control volume (this can be a very interesting feature, especially if support for speakerphone mode is provided)
- Displaying call duration (doing this right now is close to impossible)
- Quick access to the page from which the call is being made (leaving aside "media playback" notification, right now it would require a persistent notification [which is not yet supported], or having the Service Worker recreate the notification after it is dismissed or tapped)
And after the call:
- Sometimes it could be useful to have "Call back" button and/or the information about the completed call (person, duration, cost). Some reasons for this: when the call suddenly drops, the person can quickly call back; when the call has ended, the person can confirm that the call is over; when paid calls are involved (VoIP-to-PSTN for example), the cost of the call can be displayed immediately.
There are a lot of things to discuss, especially the complex use-cases. For example, what if the person receives multiple calls at once, or what if the app supports conference calls, and another participant is added to the call? Would there be a "hold" mode, where one call is left hanging, while the other call is active? What about "mute" buttons, is it a responsibility of MediaSession API? What about when both audio and video streams are present? Should they both be toggleable from the "notification"?
youennf
tomayac
In addition to supporting media playback and access to the media action keys present on many platforms the MediaSession API also seems to fit the bill for telephony use cases such as notifying the user of an incoming or active call and providing on/off hook and volume controls.