Video backends that run asynchronously can wind up in a race situation if a mode or position update request queues up a fullscreen enter request while an unprocessed asynchronous leave request is in flight, and the mode switch or position update request is processed after the leave request, causing the window to erroneously return to fullscreen.
Instead of the internal SDL_UpdateFullscreenMode and backend SetWindowFullscreen functions taking a boolean value, use an enum that allows the backends to distinguish between "this is an explicit fullscreen enter/leave request", and "this is an update request to change the mode or position". Communicating the specific intent allows the backend to early-out when required, which prevents windows from erroneously returning to fullscreen due to update requests made after a leave request, and allows for the removal of some internal synchronization previously needed to (attempt to) work around this, which improves overall performance while also increasing robustness.
This is only relevant to the internal functions, and nothing in the public-facing API has been changed.
Added SDL_Vulkan_DestroySurface, its documentation and corresponding platform specific implementations. Fixed some header inclusion orders to improve consistency between platforms. Added TODOs regarding MetalView creation and destruction which will benefit from the new functionality.
- Adds support for modal windows to Win32, Mac, and Haiku, and enhances functionality on Wayland and X11, which previous set only the parent window, but not the modal state.
- Windows can be declared modal at creation time, and the modal state can be toggled at any time via SDL_SetWindowModalFor() (tested with UE5 through sdl2-compat).
- Allows dynamic unparenting/reparenting of windows.
- Includes a modal window test.
- A refactor changed how SetupWindowData handled external windows where previously it was tracked
in a field of SDL_CocoaWindowData but now it is tracked by setting SDL_WINDOW_EXTERNAL in the
window flags. Removed the now unused field and updated the external window check in DestroyWindow.
This fixes an macOS bug that is only known to occur in fullscreen windows on the built-in displays of newer MacBooks with camera notches. When the mouse is moved near the top of such a window (within about 44 units) and then moved back down, the cursor rects aren't respected. This can cause the default cursor to be visible when it should not be.
(cherry picked from commit f1690e265e)
Alter the video driver grab/confinement function signatures to return an int, set and return an error if the grab request fails, and clear the grab flags from the window if the mouse and/or keyboard wasn't actually grabbed.
This better reflects how HDR content is actually used, e.g. most content is in the SDR range, with specular highlights and bright details beyond the SDR range, in the HDR headroom.
This more closely matches how HDR is handled on Apple platforms, as EDR.
This also greatly simplifies application code which no longer has to think about color scaling. SDR content is rendered at the appropriate brightness automatically, and HDR content is scaled to the correct range for the display HDR headroom.
Zeroing the window width and height was necessary in SDL2 to short-circuit the resize event deduplication code when the window backing scale changed, but not the logical size. This is no longer necessary in SDL3, as it will explicitly check for scale changes on resize events and dispatch pixel size/scale changed events as appropriate, even if the window's logical size hasn't changed.
- If a window being destroyed is a child of an inactive window and was the last keyboard focus of the window, that window will be left with a stale pointer
to the destroyed window that it will attempt to restore the next time that window is focused. SDL_DestroyWindow will have already taken care of moving
focus if this window is the current SDL keyboard focus so this change intentionally does not set focus.
- Like Cocoa_HideWindow, this attempts to move the focus to the closest parent window that is not hidden or destroying.
- We intentionally don't raise the application when raising a child window to allow raising a child window to the top without setting the application active but the
child window should still be set as key window for the application if desired.
- Child windows are added and removed dynamically from the window hierarchy when they're shown/hidden. Adding a hidden child window to a visible
parent is fine, but adding a hidden child window to a hidden parent will cause the child to show when the parent window is shown as it's still a part of
the window hierarchy.
- For some reason, not adding the child window to the parent entirely causes the child to not focus correctly the first time it's shown. Adding then immediately
calling orderOut to remove the child window from the hierarchy does work correctly so we do this to work around the weird issue.
For some reason, fullscreen space windows won't get any mouse button events unless the NSWindowStyleMaskTitled flag is set when entering fullscreen, even though they successfully become key and receive mouse motion events. Make sure the flag is always set when entering fullscreen space state.
Don't mark fullscreen windows as having a title bar if the window is borderless, or it can end up in a weird, pseudo-decorated state when leaving fullscreen if the borders were toggled back on while the window was fullscreen.
Setting the window styling when about to leave fullscreen caused issues with this as well and is no longer needed, as it is ensured that the window's resizable state won't change while in a fullscreen space, or in a transition period.
Don't check the fullscreen flag when toggling resizable, bordered, always on top, minimum size and maximum size, as the flag doesn't reflect pending async changes that may be in progress.
These properties can be made to be safely toggled while the window is in fullscreen mode and applied when returning to windowed mode, which ensures that requested window settings aren't lost if calling these functions while async fullscreen changes are in flight.
SDL window size, state, and position functions have been considered immediate, with their effects assuming to have taken effect upon successful return of the function. However, several windowing systems handle these requests asynchronously, resulting in the functions blocking until the changes have taken effect, potentially for long periods of time. Additionally, some windowing systems treat these as requests, and can potentially deny or fulfill the request in a manner differently than the application expects, such as not allowing a window to be positioned or sized beyond desktop borders, prohibiting fullscreen, and so on.
With these changes, applications can make requests of the window manager that do not block, with the understanding that an associated event will be sent if the request is fulfilled. Currently, size, position, maximize, minimize, and fullscreen calls are handled as asynchronous requests, with events being returned if the request is honored. If the application requires that the change take effect immediately, it can call the new SDL_SyncWindow function, which will attempt to block until the request is fulfilled, or some arbitrary timeout period elapses, the duration of which depends not only on the windowing system, but on the operation requested as well (e.g. a 100ms timeout is fine for most X11 events, but maximizing a window can take considerably longer for some reason). There is also a new hint 'SDL_VIDEO_SYNC_ALL_WINDOW_OPS' that will mimic the old behavior by synchronizing after every window operation with, again, the understanding that using this may result in the associated calls blocking for a relatively long period.
The deferred model also results in the window size and position getters not reporting false coordinates anymore, as they only forward what the window manager reports vs allowing applications to set arbitrary values, and fullscreen enter/leave events that were initiated via the window manager update the window state appropriately, where they didn't before.
Care was taken to ensure that order of operations is maintained, and that requests are not ignored or dropped. This does require some implicit internal synchronization in the various backends if many requests are made in a short period, as some state and behavior depends on other bits of state that need to be known at that particular point in time, but this isn't something that typical applications will hit, unless they are sending a lot of window state in a short time as the tests do.
The automated tests developed to test the previous behavior also resulted in previously undefined behavior being defined and normalized across platforms, particularly when it comes to the sizing and positioning of windows when they are in a fixed-size state, such as maximized or fullscreen. Size and position requests made when the window is not in a movable or resizable state will be deferred until it can be applied, so no requests are lost. These changes fix another long-standing issue with renderers recreating maximized windows, where the original non-maximized size was lost, resulting in the window being restored to the wrong size. All automated video tests pass across all platforms.
Overall, the "make a request/get an event" model better reflects how most windowing systems work, and some backends avoid spending significant time blocking while waiting for operations to complete.
This means the allocator's caller doesn't need to use SDL_OutOfMemory directly
if the allocation fails.
This applies to the usual allocators: SDL_malloc, SDL_calloc, SDL_realloc
(all of these regardless of if the app supplied a custom allocator or we're
using system malloc() or an internal copy of dlmalloc under the hood),
SDL_aligned_alloc, SDL_small_alloc, SDL_strdup, SDL_asprintf, SDL_wcsdup...
probably others. If it returns something you can pass to SDL_free, it should
work.
The caller might still need to use SDL_OutOfMemory if something that wasn't
SDL allocated the memory: operator new in C++ code, Objective-C's alloc
message, win32 GlobalAlloc, etc.
Fixes#8642.
