When a call is placed, phosh should automatically switch to the make a call configuration of audio. Without this, users make mistakenly believe that the audio calls are broken.
This behavior is supported in debian+phosh. Im currently doing research into what they are doing differently.
According to debian+phosh's README, they are compiling alot of their packages from purism's fork. That is understandable as alot of the changes have not been accepted upstream similar to geary in #459 (closed).
In relation to this issue, purism has done extensive work with Dynamically manage audio routing:
I don't know how purism and debian implemented it but I was thinking this could be implemented using feedbackd. When a call comes in, the phone-incoming-call event is triggered by feedbackd (https://source.puri.sm/Librem5/feedbackd/blob/master/Event-naming-spec-0.0.0.md). When that event is triggered we switch to "make phone call" audio profile. What do you guys think?
According to debian+phosh's README, they are compiling alot of their packages from purism's fork. That is understandable as alot of the changes have not been accepted upstream similar to geary in #459 (closed).
Well, the best solution would be to get those upstream ASAP but I guess most things would only land in GNOME 3.38 now, so in the meantime pmOS will probably have to use some of the forks (since those won't get into Alpine).
When that event is triggered we switch to "make phone call" audio profile. What do you guys think?
Hm, not sure how we'd do that switch. I guess we should update feedbackd though, I'll look into that.
According to the debian+phosh wiki, here's how they did it:
calls
Voice calls application.
Modifed to use the PinePhone's audio device name for switching to Speaker output, and replaced the "Add Call" button with a "Headset" button, switching to headphone + headset mic during a call (can be removed when jack insertion detection is implemented in driver sun50i-codec-analog, if it ever happens). None of these changes can be upstreamed.
Thanks for opening this issue, and for all the information that has been contributed.
I did not realize that we have two issues for this already. Closing this one in favor of the newer one: #644 (closed)
(Note that a MR is open to have a separate package with mobian's patched wys, until this is upstreamed properly: !1406 (merged). Please continue the discussion in #644 (closed).)