https://bugs.kde.org/show_bug.cgi?id=498362

--- Comment #4 from Kacper Paczos <[email protected]> ---
Still there on current master (plasma-pa git20260825, Plasma 6.7.80, KDE neon
unstable).

It reproduces without any capture hardware, which makes it easy to test in a
VM:

    pactl load-module module-remap-source source_name=fakemic
master=auto_null.monitor   # stands in for a microphone, stays idle
    parecord --device=auto_null.monitor --raw /dev/null                        
         # records the output

The tray item then reads "parecord is using the microphone" with a
microphone-sensitivity-high icon, while the only recording stream sits on the
monitor of a sink and the (fake) microphone is never touched. The remap-source
line is only needed because a machine with no capture device at all is caught
by the guard added in bug 410637; on real hardware the built-in mic is enough.

Cause: recordingApplications() in src/qml/microphoneindicator.cpp walks
SourceOutputModel and skips only virtual streams, nothing looks at what a
stream is attached to. PulseAudioQt drops monitors from SourceModel
(source_cb() in context.cpp returns early when monitor_of_sink is set), so a
monitor stream never matches a source there. Sinks do know their monitor's
index, so SinkModel can tell the two cases apart.

Merge request, which makes the item say "<app> is recording the audio output
(<device>)" with a different icon instead of claiming the microphone:
https://invent.kde.org/plasma/plasma-pa/-/merge_requests/423

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

Reply via email to