Allow mono input device and channel selection for stereo input devices.

[pgScorpio]

Describe the bug

Momentarily it is impossible to use a mono input device and also impossible to select input channels for stereo devices, though there seems to be no reason why this should not be possible, since it is possible to select twice the same input channel with devices with more than two input channels.
Also in some cases it would be handy to swap L/R output channels which is currently also impossible for two channel devices.

To Reproduce

Just try different devices:
mono input device: Error not 2 ch.
stereo input device: no input/output selection possible
more than 2 channel device: all is possible.

Expected behavior

Always being able to select L/R channels for input and output. (and for mono input devices always select twice the same mono input)

Screenshots

n.a.

Operating system

All OS'es

Version of Jamulus

All versions

Additional context

Also discussed on discussion #2268

changes for v3.8.1 windows release:

Add in sound.h:
(In my opinion these and some other definitions in sound.h should actually be defined in global.h since they are client, and not OS, related, but are of the same category as MAX_NUM_IN_OUT_CHANNELS which is defined in global.h)

42 #define MIN_IN_CHANNELS 1
43 #define MIN_OUT_CHANNELS 2

Chances in sound.cpp:

168 if ( ( lNumInChan < MIN_IN_CHANNELS ) || ( lNumOutChan < MIN_OUT_CHANNELS ) )

586 vSelectedInputChannels[1]  = GetNumInputChannels() > 1 ? 1 : 0;

588 vSelectedOutputChannels[1] = GetNumOutputChannels() > 1 ? 1 : 0;

Same (alike) changes also have to be done for Mac/Linux/Android sound.cpp, sound.h

Chance in clientsettingsdlg.cpp

823 if ( ( iNumInChannels < MIN_IN_CHANNELS  ) || ( iNumOutChannels < MIN_OUT_CHANNELS  ) )

I don't think there is any other impact on the GUI, but that has to be further checked/tested.

[ann0see]

Great thanks. If no discussion comes up the next 2 day, feel free to open a PR with these changes. I hope it doesn’t have an impact on other functionality (you never know…)

[henkdegroot]

should actually be defined in global.h since they are client, and not OS,

Sound.h is in the windows folder for a reason, and believe this is specifically for Windows OS only. Also the input/output selection option only exists in Windows. Linux (using JACK) is using the jack control options to select/connect and macOS is listen the different option in the "Audio Device" drop down.

EDIT: In the clientsettingsdlg.cpp, it does seem to have logic for macOS as well. I don't have a mac to test/check this. I actually wonder why these selection option should ever be hidden (apart from other OS's then windows and potentially macOS).

[pljones]

I actually wonder why these selection option should ever be hidden (apart from other OS's then windows and potentially macOS).

None of the top panel should even be in the client settings dialog -- it's all sound system specific and should be laid out there... Ah well. The channel selectors are not, and I agree - I've never understood why they get hidden under certain circumstances.

[henkdegroot]

@pljones, do you believe the channel selectors are beneficial to have on others OSs (like Linux)? Would they actually have a purpose there?

[pljones]

Historically, I'd expect the channel selectors were only added when ASIO support was added, to enable this choice. With Jack,
you'd be expected to connect correctly using qJackCtl.

But still, if I've a stereo pair connected and Jamulus decides to connect them a certain way, it's strange that if I have more than two ins, I can adjust that but if I only have two ins, I cannot, regardless of platform. I just feel I should be allowed to set it how I want.

I've actually not come across a problem myself with the way it is now -- I use Reaper to control routing and it allows me the flexibility I need. That's because it has "more than 2" inputs, so I can adjust each client I run to use the inputs I want.

The current recommended way to handle separate inputs is to run multiple instances of Jamulus. But that requires an ASIO driver (like ReaRoute) that supports this kind of set up.

[henkdegroot]

With Jack, you'd be expected to connect correctly using qJackCtl.

So we should update the logic to always show the channel selection, except when JACK is used.

@pgScorpio, you happy with that? And will you do a PR for this or do you want someone else (i.e. me) to do it?

[pgScorpio]

@henkdegroot

Sound.h is in the windows folder for a reason, and believe this is specifically for Windows OS only.

But, as far as I can see, some definitions in sound.h are client related, not sound related, and so are (and should be) equal in all sound.h files....

... and macOS is listen the different option in the "Audio Device" drop down.

And to be consistent it would be nice if we eventually could change this to the same way of input/output selection as on other OS'es too... (I personally find this list on macOS very inconvenient!)

do you believe the channel selectors are beneficial to have on others OSs (like Linux)? Would they actually have a purpose there?

I think Yes.
1: To use one input channel of a stereo device twice to have (pseudo) mono.
2: To be able to swap L/R output channels.

And even though Jack can do this, not everyone is familiar with Jack, and in my opinion we should try to have the GUI as consistent as possible for all OS'es, so preventing all kinds of conditional code with all it's risks.
And maybe, in the future, we even could consider to configure Jack from Jamulus using the input/output selection in Jamulus.

So we should update the logic to always show the channel selection, except when JACK is used.

See previous remark. So could (should?) also be valid for Jack.

you happy with that? And will you do a PR for this or do you want someone else (i.e. me) to do it?

If I can get my build environment up and running soon (for all OS'es) I will happily do the PR if you prefer that, but since it's initially just a few lines of code it would be much faster if someone more experienced with building Jamulus could already implement and test it for all OS'es.

[pgScorpio]

But still, if I've a stereo pair connected and Jamulus decides to connect them a certain way, it's strange that if I have more than two ins, I can adjust that but if I only have two ins, I cannot, regardless of platform. I just feel I should be allowed to set it how I want.

That's exactly my point of view...

[henkdegroot]

I have done some testing and looks like we should improve the "spacing" in the input/output selection section:

As you can see in my screenshot, the important part (using input 1 or 2/output 1 or 2) is not visible. In the pull down it is possible to see (but that is because text contains some ... to make it fit.

I don't think there will be any translation which increases the space required for the L and R text.

For JACK, I don't see any added value with the current options as it just lists two options named "Default":

[pgScorpio]

For JACK, I don't see any added value with the current options as it just lists two options named "Default":

Where does this "Default" come from?? It makes no sense to me. "Default" could be in the selection list, but the box should always show the actual selected input/output.

Also I think for jack it should still be possible to swap Left and Richt channels... Just for consistency.
Please keep the UI the same as much as possible for all platforms.

(As soon as I'm up and running to build all OS Jamulus versions I will start working on a platform independent sound interface, and so the UI and the way of selecting inputs/outputs will be the same for all platforms too.)

[ann0see]

As soon as I'm up and running to build all OS Jamulus versions I will start working on a platform independent sound interface, and so the UI and the way of selecting inputs/outputs will be the same for all platforms too.

Although unrelated, it would be great if you document your work in the docs folder as a short overview