InputSource, replaces #558

[UE4SS]

This PR is rebased onto main.
The old one is a disaster of a PR, this one contains only the InputSource commits and is rebased onto main.
Please see #558 for details.

This PR needs testing.
Keybinds seem to work fine for me on Linux using Proton, tested with Palworld.
Someone needs to test on Windows natively.

[github-actions]

Build Details

Name	Information
PR Commit	aedd3f1
Merge Commit	2c55c2e
Size	46.93 MB
Last Updated	Mar 12, 25, 2:21:52 PM UTC
Expires At	Mar 26, 25, 2:21:47 PM UTC

---

Build Details

Name	Information
PR Commit	aedd3f1
Merge Commit	2c55c2e
Size	28.59 MB
Last Updated	Mar 12, 25, 2:25:47 PM UTC
Expires At	Mar 26, 25, 2:25:44 PM UTC

[narknon]

How crazy would it be to make the register_keydown_event and other functions/lambdas just compile into nothing rather than wrapping every use of them in ifdefs?

[UE4SS]

How crazy would it be to make the register_keydown_event and other functions/lambdas just compile into nothing rather than wrapping every use of them in ifdefs?

I agree.
I think it's better to wrap the contents of the functions with #ifdef instead so that it's empty when not supported.

[UE4SS]

How crazy would it be to make the register_keydown_event and other functions/lambdas just compile into nothing rather than wrapping every use of them in ifdefs?

I agree. I think it's better to wrap the contents of the functions with #ifdef instead so that it's empty when not supported.

Although this could be tricky with functions that use a type that is also #ifdef'd out, like get_input_handler.
You'd have to dummy that type to make it compile, and then it also returns a reference which means there needs to be an #else that returns a static version of the dummy type.
It makes it more complicated to implement but removes all burden from users of the API which I think is better.

Note that get_input_handler doesn't seem to be used anywhere, and doesn't exist before this PR so it should probably be removed.

[narknon]

Can you just ifdef the members?

[UE4SS]

Can you just ifdef the members?

I don't see how that would help here.
If the members are ifdef'd out, anything that uses those members won't compile.

[narknon]

Won't the only things using the members be the ifdef'd functions? Idk, I haven't looked closely at it.

[UE4SS]

Won't the only things using the members be the ifdef'd functions? Idk, I haven't looked closely at it.

Members and functions are currently fully ifdef'd out.
As a result, users of those members and functions also need an ifdef around them or they won't compile.
This would be inconvenient for mods that register keybinds for example.
They shouldn't have to ifdef their own calls to the registration function.
If we remove get_input_handler, I think the registration functions are the only place where this matters as everything else is internal stuff.

[UE4SS]

The discussion above has been resolved, the current status is now:

1. Waiting for someone to test and confirm that keybinds work as expected on native Windows.
   There are build artifacts above, you don't have to build it yourself.
2. Waiting for @Yangff to possibly have a quick look.

[igromanru]

I tested the latest build from here with my mods, mainly with my StackManage mod, which utilize multiple RegisterKeyBind with same primary key and different modifiers.
https://github.com/igromanru/StackManager-UE4SS-AF/blob/2a0c2471152805f449aeb34c22298eafccf9c662/Scripts/main.lua#L12
Couldn't find any issues.

[Yangff]

hmm.. it looks good to me.

[UE4SS]

Approval is pending by @narknon, and it's ready to be merged immediately after.

[narknon]

If you don't mind, please manually squash the last two commits then do a normal merge. Not a huge deal though.