Add MIDI controller device index to InputEventMIDI.device property.

[fbcosentino]

It is possible to query the OS for the connected MIDI controllers, but the event messages' device field was not being used. This implements controller index being sent in InputEventMIDI messages in the device property, matching the index from OS.get_connected_midi_inputs().

Based on the work done by @ramdor in this archived PR

Closes godotengine/godot-proposals#7733

[AThousandShips]

The co-author didn't apply, you need to make the part in the arrow brackets their email, check out their branch and use git log to get a valid author to use 🙂

[fbcosentino]

The co-author didn't apply, you need to make the part in the arrow brackets their email, check out their branch and use git log to get a valid author to use 🙂

Done. Seems to work now.

[AThousandShips]

Your commit seems not to be linked to your GitHub account. See: Why are my commits linked to the wrong user? for more info.

[fbcosentino]

Your commit seems not to be linked to your GitHub account. See: Why are my commits linked to the wrong user? for more info.

Fixed. (Apparently I'm as good with github as Sheldon is good with the flute. Eventually I'll get better.)

[PeterMarques]

This needs to get pushed up.

[Sauermann]

I have no experience with drivers, but the changes to core are looking correct.

[Sauermann]

Changes to core are looking correct.
Can't say anything about the change to drivers.
I can't test the PR, because I don't have MIDI-devices.

[ramdor]

hi @fbcosentino . I know this has been merged now, based on the changes I was dabbling with a long time ago. However, there are some issues with the coremidi implementation. I built a quick mac test and it seems like the device_index (InputEvent.device) is garbage data.

It looks to be caused by line 68 of midi_drive_coremidi.cpp where the pointer to 'i' is being used. This should not be a pointer, but the actual value of 'i'. Otherwise multiple read callbacks will use the same pointer. Also 'i' will go out of scope and result in junk.

We need to provide the actual value of i to MIDIPortConnectSource so that when the callback function read is called when midi events come in, src_conn_ref_con is the actual value of i that was provided when MIDIPortConnectSource was called and can be used as device_index.

I hope this makes sense. Cheers.

[ramdor]

some test code that does correct type casting :

#include <iostream>
//#include <cstdint>

void testFunction(void *src_conn_ref_con)
{
    intptr_t device_index = reinterpret_cast<intptr_t>(src_conn_ref_con);
    std::cout << device_index << std::endl;
}

int main()
{
    int sources = 16;
    for (int i = 0; i < sources; i++) {
        testFunction(reinterpret_cast<void*>(static_cast<intptr_t>(i)));
    }
}

[fbcosentino]

some test code that does correct type casting :

#include <iostream>
//#include <cstdint>

void testFunction(void *src_conn_ref_con)
{
    intptr_t device_index = reinterpret_cast<intptr_t>(src_conn_ref_con);
    std::cout << device_index << std::endl;
}

int main()
{
    int sources = 16;
    for (int i = 0; i < sources; i++) {
        testFunction(reinterpret_cast<void*>(static_cast<intptr_t>(i)));
    }
}

Hi @ramdor !

I actually don't have a mac (nor access to one). Would you be comfortable to make these changes yourself?

[ramdor]

@fbcosentino yes, i have just started a PR. I couldn't test it either, but it should be ok.