This bug report was migrated from our old Bugzilla tracker.

These attachments are available in the static archive:

• Test cases (sdl2_minimum_size_bug.c, text/x-csrc, 2014-03-18 21:05:07 +0000, 3936 bytes)

Reported in version: HG 2.0
Reported for operating system, platform: Linux, x86_64

Comments on the original bug report:

On 2014-03-18 21:05:07 +0000, wrote:

Created attachment 1594
Test cases

A window can be resized to out-ot-bounds dimensions even if any of SDL_SetWindowMinimumSize and SDL_SetWindowMaximumSize is called, given the following steps are done in the given order:

• First, the window is created without the SDL_WINDOW_OPENGL flag set.
• Then SDL_SetWindowMinimumSize and/or SDL_SetWindowMaximumSize are/is called.
• Afterwards a new renderer is created.

Sample code is attached with a test case. By playing with a few definitions given at the top of the code you can generate a few more tests:

• Manage a GL context directly (SDL_GL_CreateContext) or access an SDL 1.2 style window surface (SDL_GetWindowSurface), rather than using a renderer (SDL_CreateRenderer).
• Set window bounds after calling any of these (say SDL_CreateRenderer) rather than before.
• Optionally add SDL_WINDOW_OPENGL as a flag on window creation.

A few gotchas that should be added for the various cases:

• An OpenGL context cannot be created without SDL_WINDOW_OPENGL. I have added a compile-time error shown upon an attempt to do so.
• If the window bounds are set after the call to SDL_GetWindowSurface, then the surface gets invalidated and a new surface should be retrieved again (I suppose that's the expected behavior).
• As in the case of SDL_CreateRenderer, if the bounds are set before the (only) call to SDL_GetWindowSurface then the window can be resized to out-of-bounds dimensions.

In addition:

• If the hint SDL_HINT_RENDER_DRIVER is set to "software" before the call to SDL_Hint is done (say by setting the environment variable SDL_RENDER_DRIVER to the same value), then the bounds hold (with the same code as originally attached to this bug report).
• The latter point applies if a window surface is used, too.

On 2015-02-19 05:22:22 +0000, Ryan C. Gordon wrote:

Marking a large number of bugs with the "triage-2.0.4" keyword at once. Sorry if you got a lot of email from this. This is to help me sort through some bugs in regards to a 2.0.4 release. We may or may not fix this bug for 2.0.4, though!

On 2016-01-31 14:27:08 +0000, Rafał Mużyło wrote:

I think I see a related problem.

This seems to be a case of unintended side-effects, that are beneficial here.

SDL_WINDOW_OPENGL is the flag that has those side-effects.

Another test would be to create a FULLSCREEN_DESKTOP window smaller than the desktop.

If you unset fullscreen, the window will still have the size of the desktop.

If SDL_WINDOW_OPENGL is set, once again, the window is restored to its expected size.