[AD] Various tangents to distinguishing keyboards

["S. J. Dobrowolski" <stan423@xxxxxxxxx>]

Grettings, everyone.

Apologies if this is the wrong channel but it relates to source code of 
Allegro so I couldn't come up with anything more appropriate. (Also, 
please excuse any English errors.) I was considering modifying A5 for 
multiple mice and keyboards event input support - while distinguishing 
these devices, for clarity. Rationale: some of my friends and me don't 
like gamepads that much, but like the idea of local play with separate 
controllers. As a sidenote, I guess I'm more familiar with using A4, but 
my prototypes could use a rewrite, even if it means coding from scratch 
where I have used A4 bitmaps for game logic.

Based solely on web searches, on Windows, it appears some sort of 
hardware ID is part of "raw input" API event. On Linux, it appears that 
evdev API allows you to distinguish connected devices, but I'm not sure 
how that is supposed to cooperate with various layers of user protection 
and desktop stack. On MacOS... I frankly couldn't find good information. 
I don't have a Mac anyway, so trusting my Mac code even in the slightest 
would be a huge mistake.

Based solely on my glancing at Github Allegro 5 code, the Windows 
version seems to be using the "regular" window events for keyboard 
and/or mouse handling instead of "raw input". However there is a "raw 
input" handler in what looks like the window event queue handler; 
comment says it is unused though. It doesn't appear to handle anything 
but mouse events, and does not appear to try to distinguish the mice. 
The Linux version appears to be using evdev for mice and joysticks... 
well, the mouse init code made me rewrite this paragraph multiple times, 
is that supposed to be a "mixed" mouse or just a single one, and how do 
we know it's that early in the file system when joysticks need to check 
such a wider range? I guess I should install Linux and see for myself. 
Maybe no Linux Allegro users happened to be Linux Allegro laptop users 
who plugged in a regular mouse, maybe I'm missing something monumental 
about evdev.

So I guess I'm writing to you with following questions:

* Would you consider "multikeyboard" and "multimouse" add-ons to be 
added into the mainstream version of A5? Would real access to the 
feature (as opposed to an API wrapper) being limited to Win/Lin be a 
problem? Would you be concerned about this ending up being unmaintained?

* Does that sound like an overkill project for a team project beginner?

* Is no veteran maintainer doing something like this quietly at the moment?

* Can I assume that opening more evdev descriptors on Linux and doing 
stuff with them, as well as requesting both types of input events on 
Windows and then handling the new one has no good reason to break 
existing input framework?

* Can I just write what I want in Windows raw input handler and delete 
the code marked as dead, or is that there for a reason? Did someone try 
in the past to implement a raw input driver and ended up with loads of 
problems?

* What does Linux mouse driver init actually do?

* Is there anything that sounds like an obvious problem already?

Best regards,

Stan