Bluetooth mouse not detected

[HerrBvomSee]

I am restarting gamedev with Allegro5 on a new iMac and use primarily a bluetooth mouse (Logitech MX Master) which is working smoothly in all apps.
When I start my self written game, I call al_install_mouse() which fails, as it seems that it can not detect the mouse properly.
Attaching a USB mouse helps, but I do not want to have multiple mice connected to my machine just for development purposes.
The init order is

al_init()
al_install_keyboard()  

and then

al_install_mouse()

, which returns all the time with false, even with the properly (in terms of system usage) BT mouse attached.

The allegro.log returns

system I system.c:257 al_install_system [ 0.00000] Allegro version: 5.2.3 (GIT)
MacOSX I qzmouse.m:236 osx_init_mouse [ 0.00028] Detected 0 pointing devices
MacOSX I qzmouse.m:271 osx_init_mouse [ 0.00029] Device effectively has 0 axes and 0

[HerrBvomSee]

Generated a patch for the issue - working at least on my machine. If this patch seems helpful, please review.
mouse_detect_patch.diff

[ghost]

Thanks for the patch. I took a quick look and it looks OK (for my reference did you change just the top of the big code block, the patch is a little wonky...?) I'll take a look at applying it soon.

[HerrBvomSee]

@goobliata thanks for your feedback. I add here the info I mentioned in the IRC related to the patch:

At first, kHIDUsage_GD_Mouse did not detect the bluetooth mouse. Using the kHIDUsage_GD_Pointer instead detects it (as it is a "lower level" pointing device (from what I read - can not find it anymore :( ... ))

To not break the old code, I thought to test on both types, at first the GD_Mouse to keep the standard "as-is". If this does not detect any devices, I will use the fallback of GD_pointer.
To use both searches I introduced a new function _get_matching_services to get the iterator

Also I changed the dictionary used for the search from kIOHIDPrimaryUsageKey and kIOHIDPrimaryUsagePageKey to kIOHIDDeviceUsageKey and kIOHIDDeviceUsagePageKey as the former ones only take ONE usage into account, so when you have a keyboard with mousepad, it might detect only the keyboard and neglects the mouse part, resulting in no mouse
so I hope the code does not break old stuff and just added an additional feature/fix :)

ah and yes, the upper part is changed only. The rest are just formatting changes, as the idention level has changed due to new function.

And to improve myself in generating/writing patches: what do you mean with wonky?

[ghost]

Thanks for the clarification. I don't have a bluetooth mouse but I'll test with a USB mouse and a keyboard/trackpad combo. But by the look of it it hopefully shouldn't break compatibility.

[ghost]

There's a pull request here if anybody wants to test it with a Bluetooth mouse before it gets committed #748

[ghost]

Patch has been applied. Thanks!