[rcore][RGFW] bug fixes

[ColleagueRiley]

• ensure functions that use platform.window check platform.window is valid first
• ensure fullscreen toggling is working properly
• ensure raylib responds to RGFW_quit properly
• update RGFW
• handle RGFW_windowMaximized, RGFW_windowMinimized and RGFW_windowRestored events
• properly handle raylib's FLAG_WINDOW_UNFOCUSED flag with RGFW_windowFocusOnShow
• remove out-of-date code

[sleeptightAnsiC]

ensure functions that use platform.window check platform.window is valid first

This is something similar that rcore_desktop_glfw.c does inside of WindowShouldClose() but no other function has those kind of checks. Other backends also don't have them. If this gets accepted, can we also port it to SDL, GLFW, etc? @raysan5

(also linking this to #4751 )

[raysan5]

@ColleagueRiley @sleeptightAnsiC In raylib I tried to avoid/minimize this kind of checks, spreaded all around the code on all functions. Can it be checked only on InitWindow()?

[sleeptightAnsiC]

Can it be checked only on InitWindow()?

It should be possible to early return from InitWindow() and then someone could simply check for the errors with IsWindowReady() and abort/recover accordingly.

Currently, there is a problem with raylib's initialization where, if a platform cannot provide Window and/or OpenGL context, the whole application will crash deep inside of rcore backend code, without any possibility to recover. I wanted to report/patch this, but then I noticed a similar PR #4164 where you rejected the fix @raysan5 Would it be fine, if I try to make another Issue and PR for this problem? (EDIT: I just reported the mentioned problem here: #4801 )

[raysan5]

Currently, there is a problem with raylib's initialization where, if a platform cannot provide Window and/or OpenGL context, the whole application will crash deep inside of rcore backend code, without any possibility to recover.

@sleeptightAnsiC Could it be just validated if the process (window+context creation) worked properly and just show a TraceLog() message showing the issue (like most raylib functions approach) if not initialized and then just set CORE.Window.ready to false for further user check?

[sleeptightAnsiC]

@sleeptightAnsiC Could it be ...

Replied under linked issue.

[ColleagueRiley]

If the window is not valid or is invalidated, there will be a confusing crash.

This check is there to avoid the crash, but I agree that this is not how it should be done. Instead of the if-statement, there should be an assert on each function platform.window so they crash in an obvious way if they are used before initialization or after deinitialization.

If you think there shouldn't be any asserts or checks, I will remove them. But they could be useful if something goes wrong.

That other issue is not relevant for RGFW internally, RGFW fallbacks to a basic OpenGL context if it fails to load the accelerated context. The crash will still happen because RLGL does not use a fallback, but that could be easily added.

As for SetWindowState: It assumes that the user might try setting the state before the window is created and modifies flags that can only take effect when creating the window. that may not be a correct assumption.

[raysan5]

I prefer to avoid adding that check to all the functions, if that check is required I think it should be implemented in RGFW library side.

[ColleagueRiley]

Alright, they exist on the RGFW side as asserts.

[raysan5]

@ColleagueRiley thanks for the review!