https://bugs.kde.org/show_bug.cgi?id=518378

Paul Gratz <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[email protected]

--- Comment #1 from Paul Gratz <[email protected]> ---

I hit this same bug (Ubuntu 26.04, Plasma 6.6.6, plasma-browser-integration
6.6.4,
extension 2.2, Google Chrome 154) and traced it to the media-controls hack in
page-script.js, addPlayerToDomEvadingAutoPlayBlocking().

Google Voice creates its ringback/ringtone as `new Audio()` with loop=true and
calls play(). The extension's "play" listener (registerInDom) appends the
element
to <head> and immediately removes it. Per the HTML spec, removing a media
element
from the document pauses it, so a "pause" event fires. replayAfterRemoval then
calls play() again.

Google Voice listens for that "pause" event on its own element and treats it as
"ringback stopped", so it drops its reference. When the call connects or ends,
Voice never calls pause() on the element again. The element is now orphaned,
not
in the DOM, looping forever. I instrumented HTMLMediaElement.prototype in the
tab
and confirmed: after play() there is zero pause()/src/loop/remove activity from
Voice for the rest of the call, only the extension's removeChild + play().
Calling
pause() on the orphaned element by hand stops the sound immediately.

Side note: line ~180 has a typo, `player.replyAfterRemoval` instead of
`replayAfterRemoval`, so the pause listener is never actually removed.

Workaround: add https://voice.google.com to "Block media controls on the
following
domains" in the extension options and reload the tab. This fixes it completely.

Suggested fix: add "https://voice.google.com": false to MPRIS_WEBSITE_SETTINGS
in
constants.js alongside meet.google.com, zoom.us and teams.microsoft.com, which
are
already excluded for the same reason. This likely also fixes bug 472919, since
Voice's "browser might prevent calls from ringing" banner is probably triggered
by
the same spurious pause during its autoplay probe.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to