Why are KEY_* constants not used in the DRM backend?

[matthijskooijman]

Looking over the code, I noticed a lot of keycode literals, for example:

raylib/src/platforms/rcore_drm.c

Lines 176 to 182 in 852c513

	static const short linuxToRaylibMap[KEYMAP_SIZE] = {
	// We don't map those with designated initialization, because we would getting
	// into loads of naming conflicts
	0, 256, 49, 50, 51, 52, 53, 54,
	55, 56, 57, 48, 45, 61, 259, 258,
	81, 87, 69, 82, 84, 89, 85, 73,
	79, 80, 91, 93, 257, 341, 65, 83,

and

raylib/src/platforms/rcore_drm.c

Lines 1512 to 1515 in 852c513

	case 0x41: CORE.Input.Keyboard.currentKeyState[290] = 1; break; // raylib KEY_F1
	case 0x42: CORE.Input.Keyboard.currentKeyState[291] = 1; break; // raylib KEY_F2
	case 0x43: CORE.Input.Keyboard.currentKeyState[292] = 1; break; // raylib KEY_F3
	case 0x44: CORE.Input.Keyboard.currentKeyState[293] = 1; break; // raylib KEY_F4

These values have KEY_* constants in raylib.h:

raylib/src/raylib.h

Lines 579 to 584 in 852c513

	KEY_APOSTROPHE = 39, // Key: '
	KEY_COMMA = 44, // Key: ,
	KEY_MINUS = 45, // Key: -
	KEY_PERIOD = 46, // Key: .
	KEY_SLASH = 47, // Key: /
	KEY_ZERO = 48, // Key: 0

In some places, the constant names are used in comments, but why are they not just used in place of the literals? That would make the code a lot more readable and maintainable, I would say?

[raysan5]

@matthijskooijman I think it was because it could confuse developers with the mapping and also for consistency.

[orcmid]

Yes it's very tricky, involving the character-set of the source file, the understood mapping in the compiler, and then all manner of run-time considerations around keyboard key positions versus their keytops.

I don't know that the raylib approach is optimal, but I can't imagine what it takes to do it better. One could expend a career on just this and internationalization around character code points/glyphs, and sorting order. I know someone who did that. Uncertain what happened to him.