feat: add ability to set source for backlight / brightness and volume controls

[philmoz]

Set a Pot, Slider or Switch to control brightness without using a special / global function.

If a Backlight special / global is defined it will take precedence over this new radio setting.

TODO:

• limit selection to Pots, Sliders or Switches Companion radio settings editor.

[philmoz]

@elecpower When you have some time could you look at the companion side please.
I have implemented the YAML read/write and added the control to the radio settings edit page.

It needs to be updated to limit the backlight control to only select Pots, Sliders or Switches,

[3djc]

I think this does go against the whole logic, where things are done throuht SF. Now it is a settings, a GF, a SF, for people to understand what happens , it becomes a nightware.

What happens if you usually don't need to use the pot, but this new model needs it for a gyro gain for example ?

[pfeerick]

Looks like #2256 won't be needed anymore then, and the issues it addressed can be moved to here...

[pfeerick]

I think this does go against the whole logic, where things are done throuht SF. Now it is a settings, a GF, a SF, for people to understand what happens , it becomes a nightware.

No, it gives a mechanism to it without using GF or SF. If you need something more advanced, you use that instead.

What happens if you usually don't need to use the pot, but this new model needs it for a gyro gain for example?

Same could be said for Brightness GF... i.e. you wouldn't configure it globally, or you use it for both.

[elecpower]

The commit above fixes the source list. The dynamic update part is a wip

[pitts-mo]

This PR #6684 is interesting and could be helpful to band-aid the primary problems in SF backlight I observe since introduced in EdgeTX ~v2.6.x. But this PR still does not provide a complete solution to allow setting both an alternate backlight brightness and control... Please consider/investigate issue #5851 as well.

Thank you & Happy to test :-)

[pitts-mo]

As I poke with this a bit I find myself agreed with @3djc. We are not talking about a menu navigation button, this is normal user controls and they should be left for user to configure in SF / GF.

[philmoz]

The commit above fixes the source list. The dynamic update part is a wip

Thanks for that. A couple of things:

• Tilt X & Tilt Y should not be selectable
• Negative / inverted values should be selectable.

[elecpower]

• Tilt X & Tilt Y should not be selectable
• Negative / inverted values should be selectable.

Noted and will incorporate

[elecpower]

Custom function switches not listed correctly and will be addressed as part of the larger issue

[elecpower]

TODO: add flex switches to list
Update: done

[elecpower]

@philmoz back to you for testing

[philmoz]

@philmoz back to you for testing

Looks good - thank you.

[3djc]

For consistency then, shouldn't the same apply to audio and variometer volumes ?

[philmoz]

For consistency then, shouldn't the same apply to audio and variometer volumes ?

One step at a time.

[philmoz]

Added audio volume control as well.

[elecpower]

Looks like I need to rename the source item model to a generic name. I'm open to suggestions.

[philmoz]

Looks like I need to rename the source item model to a generic name. I'm open to suggestions.

ControlSourceItemModel

[elecpower]

Testing in Ubuntu and results are inconsistent.

Used the latest PR state and setting nothing in Companion.
Set radio settings backlight no warning is displayed as expected.
Set a GF backlight and enable. Then return to radio settings backlight and no warning when there was in earlier test. Aaarrrgh.
Leave GF and add SF to the active model. Then return to radio settings backlight and warning displayed which it should be.
Deleted SF from model and warning not displayed as expected.

By the way the text is overlapping.

[elecpower]

Simulator and Companion crash after pressing Select model on any other model from Manage Models. All were created in Companion with defaults.
Stops in datastructs_screen.h line 67

[elecpower]

In Companion setting backlight source and a backlight GF.
Start simulator and go to Radio Settings and no warning. Same as performing both tasks on sim. But reverse of first tests.

[elecpower]

In the sim, if I add a second GF the warning displays and after I delete the second GF it disappears.

[philmoz]

@elecpower can you send your .etx file please.

[elecpower]

This is from the terminal window

25050ms: LONG PRESS[0x7d7180100760]
25350ms: TE_RELEASED
26250ms: TE_PRESSED
26400ms: TE_RELEASED
26400ms: CHILD_DELETED tile[0]
free(): invalid pointer

Created a new models and settings file. Added 3 models with defaults. Saved to this file
simcrashonswitchmodel.zip

Just change the extension back to etx.

[elecpower]

Backlight settings display post creation of GF

[elecpower]

[elecpower]

Closing Companion also causes a crash at same location in the libsim

900900ms: CHILD_DELETED tile[0]
900900ms: luaClose 0x7466580f0568
900900ms: luaClose 0x7466580eeed8
900900ms: sdDone
900900ms: stopped
900900ms: <timer_queue> stopped
[D] SimulatorWidget::deleteTempData():393 - Deleting temporary settings directory "/tmp/etx-lRisPx"
[D] SimulatorLoader::findSimulatorByName():102 - [simulator.interface.loader] searching "edgetx-tx15" simulator
[D] SimulatorLoader::unloadSimulator():171 - [simulator.interface.loader] Unloading "edgetx-tx15" ( "/src/build-e30-q690-debug/native/libedgetx-tx15-simulator.so" ) result: true
[D] main():270 - COMPANION EXIT 0
free(): invalid pointer

[philmoz]

The warning is only shown if the SF/GF override is active, not just defined.

[philmoz]

This is from the terminal window

25050ms: LONG PRESS[0x7d7180100760]
25350ms: TE_RELEASED
26250ms: TE_PRESSED
26400ms: TE_RELEASED
26400ms: CHILD_DELETED tile[0]
free(): invalid pointer

Created a new models and settings file. Added 3 models with defaults. Saved to this file simcrashonswitchmodel.zip

Just change the extension back to etx.

Hmmm. No crash on MacOS.

[elecpower]

The warning is only shown if the SF/GF override is active, not just defined.

So as it is a runtime conditional check there is no point me adding a warning to Companion.

Dumb question time. Why bother with a warning at all. If I set backlight in radio settings and months later add a backlight SF/GF then unless I consciously go into backlight settings and the SF/GF is active I wont know the radio setting is being overridden?

We don't display a warning in Outputs when an Override Channel SF is enabled. There are likely more stupid setting combinations that can be configured to effectively override other settings.

A comment that has been around since OTX is that there is often not just one way of achieving the same result.

[elecpower]

If it helps here is the Ubuntu call stack at the point of the crash

[pfeerick]

The warning is only shown if the SF/GF override is active, not just defined.

So as it is a runtime conditional check there is no point me adding a warning to Companion.

"Enabled" might be the better word - if the SF/GF is defined, but disabled, no warning. Defined and enabled, then warning.

Dumb question time. Why bother with a warning at all. If I set backlight in radio settings and months later add a backlight SF/GF then unless I consciously go into backlight settings and the SF/GF is active I wont know the radio setting is being overridden?

We don't display a warning in Outputs when an Override Channel SF is enabled.

We do on the channel monitor (and outputs edit screen on colorlcd) though... there is an icon shown on both B&W and colourlcd if channel override is in effect for a given channel...

If it weren't for the fact you can do things multiple ways, a warning wouldn't be necessary, to prompt you that it is really configured somewhere else...

[elecpower]

Ok so accept your points.

The code to the effort of checking for active backlight. Why not display no message when no GF/SF actions, 'Overridden by GF/SF" when active and "Overrides in GF/SF" when inactive?

[pfeerick]

I have no real opinion either way, but would suggest it adds clutter to show something when SF/GF is disabled (and thus effectively non-existent). I'll leave that for you, Phil and co. to figure out. 🤭

[elecpower]

I just don't like the idea that a SF/GF is lurking in the shadows with no indicator.

[philmoz]

If it helps here is the Ubuntu call stack at the point of the crash

That is really weird the WidgetOptionValue is a very simple structure with a small union and a std::string.

[pfeerick]

I'm seeing the same on Linux Mint here, but not on MacOS or Windows.

3060ms: sdDone
3060ms: <menus> stopped
3060ms: <timer_queue> stopped
[D] SimulatorWidget::deleteTempData():393 - Deleting temporary settings directory "/tmp/etx-PMHcBv"
[D] SimulatorLoader::findSimulatorByName():102 - [simulator.interface.loader] searching "edgetx-tx15" simulator
[D] SimulatorLoader::unloadSimulator():171 - [simulator.interface.loader] Unloading "edgetx-tx15" ( "/tmp/.mount_EdgeTXIjkDHF/usr/lib/companion30/[libedgetx-tx15-simulator.so](http://libedgetx-tx15-simulator.so/)" ) result: true
[D] main():270 - COMPANION EXIT 0
free(): invalid pointer
Aborted (core dumped)

---
gdb

[D] main():270 - COMPANION EXIT 0
[Thread 0x7ffff0bfe6c0 (LWP 23223) exited]
free(): invalid pointer

Thread 1 "AppRun" received signal SIGABRT, Aborted.
__pthread_kill_implementation (no_tid=0, signo=6, threadid=<optimised out>) at ./nptl/pthread_kill.c:44
warning: 44     ./nptl/pthread_kill.c: No such file or directory
(gdb) backtrace
#0  __pthread_kill_implementation (no_tid=0, signo=6, threadid=<optimised out>) at ./nptl/pthread_kill.c:44
#1  __pthread_kill_internal (signo=6, threadid=<optimised out>) at ./nptl/pthread_kill.c:78
#2  __GI___pthread_kill (threadid=<optimised out>, signo=signo@entry=6) at ./nptl/pthread_kill.c:89
#3  0x00007ffff544527e in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#4  0x00007ffff54288ff in __GI_abort () at ./stdlib/abort.c:79
#5  0x00007ffff54297b6 in __libc_message_impl (fmt=fmt@entry=0x7ffff55ce8d7 "%s\n")
    at ../sysdeps/posix/libc_fatal.c:134
#6  0x00007ffff54a8ff5 in malloc_printerr (str=str@entry=0x7ffff55cc672 "free(): invalid pointer")
    at ./malloc/malloc.c:5772
#7  0x00007ffff54ab38c in _int_free (av=<optimised out>, p=<optimised out>, have_lock=0) at ./malloc/malloc.c:4507
#8  0x00007ffff54addae in __GI___libc_free (mem=0x7fffca207fc8 <DateTimeWidget::options+40>) at ./malloc/malloc.c:3398
#9  0x00007fffc9d15c96 in TopBarPersistentData::~TopBarPersistentData() ()
   from /tmp/.mount_EdgeTXDBfKLL/usr/lib/companion30/[libedgetx-tx15-simulator.so](http://libedgetx-tx15-simulator.so/)
#10 0x00007ffff5447a76 in __run_exit_handlers (status=0, listp=<optimised out>,
    run_list_atexit=run_list_atexit@entry=true, run_dtors=run_dtors@entry=true) at ./stdlib/exit.c:108
#11 0x00007ffff5447bbe in __GI_exit (status=<optimised out>) at ./stdlib/exit.c:138
#12 0x00007ffff542a1d1 in __libc_start_call_main (main=main@entry=0x5555556dc210 <main>, argc=argc@entry=1,
    argv=argv@entry=0x7fffffffd778) at ../sysdeps/nptl/libc_start_call_main.h:74
#13 0x00007ffff542a28b in __libc_start_main_impl (main=0x5555556dc210 <main>, argc=1, argv=0x7fffffffd778,
    init=<optimised out>, fini=<optimised out>, rtld_fini=<optimised out>, stack_end=0x7fffffffd768)
    at ../csu/libc-start.c:360
#14 0x0000555555700d25 in _start ()

[philmoz]

Are there current & working instruction for setting up build on Ubuntu?
I have VM images for 20, 22 and 24 - so instructions for any of those would be appreciated.

[pfeerick]

Think the only build instructions are for ubuntu 20.04 and haven't been updated since...

I would just install docker so can use a variation of docker run --name="ETXDocker" -it --rm --mount src="$(pwd)",target="/src",type=bind ghcr.io/edgetx/edgetx-dev bash to just download and use the docker build container. Otherwise, might be able to reverse engineer https://github.com/EdgeTX/build-edgetx/blob/main/dev/Dockerfile enough to get it working on 22.04 native.

[philmoz]

Ubuntu 20 won't work as it has no Qt6.
There are setup scripts for 22 and 24; but neither works.
Only Docker instructions are for Windows 10 / GWSL (currently trying this; but having to reboot into bootcamp is a pain).

[philmoz]

Crash should be fixed by PR #6825. I will rebase this when that PR is merged.