The projection and view matrices are now computed ahead of time, as they both get computed in the same spot, and typically not often. If this does, however, become a performance problem later on, this change can always be reverted.
Previously, the shaders would get compiled separately, the output of which would need to be packaged into the app. This change should make SDL's dll be the only binary needed to include SDL in a WinRT app.
Due to the new "tap and hold" IME in Mountain Lion and above, we were getting
inconsistent repeat of SDL_TEXTINPUT events. Disabling that functionality (since
you can't see the popover anyway) solves this.
Bug: https://bugzilla.libsdl.org/show_bug.cgi?id=2387
The DualShock 4 has all elements listed twice: once in the top-level list of
elements, and once in an "Application Collection" element at the top-level.
Each element has a proper cookie with a unique value, so now we descend into
each element collections, but before we add an element to the device's list,
we make sure we don't already have one with that cookie, probably from
another collection or a buggy device.
Tiemo Jung
All CPU detection functions SDL_Has* will return false, even if it is supported by the CPU, if SDL is compiled with MSVC and the target is x64.
The reason for this is that 'CPU_haveCPUID' will return 0 and macro 'cpuid' in SDL_cpuinfo.c is the fallback implementation, which sets all params to zero.
It is safe to assume that cpuid is supported on a CPU that runs windows x64, so CPU_haveCPUID can just return 1, and the empty macro can be replaced with a small wrap around the __cpuid intrinsic.
Felix Geyer
Starting from version 2.0.1 libsdl fails to build on Debian GNU/kFreeBSD in SDL_cpuinfo.c.
GNU/kFreeBSD defines __FreeBSD_kernel__ but not __FreeBSD__.
The #ifdef __FreeBSD__ check should be extended for __FreeBSD_kernel__, see the attached patch.
Build log:
libtool: compile: gcc -g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -D_FORTIFY_SOURCE=2 -DUSING_GENERATED_CONFIG_H -Iinclude -I/?BUILDDIR?/libsdl2-2.0.1+dfsg1/include -mmmx -m3dnow -msse -msse2 -fvisibility=hidden -D_REENTRANT -I/usr/include/dbus-1.0 -I/usr/lib/x86_64-kfreebsd-gnu/dbus-1.0/include -DHAVE_USBHID_H -DUSBHID_NEW -D_REENTRANT -Wall -c /?BUILDDIR?/libsdl2-2.0.1+dfsg1/src/cpuinfo/SDL_cpuinfo.c -fPIC -DPIC -o build/.libs/SDL_cpuinfo.o
/?BUILDDIR?/libsdl2-2.0.1+dfsg1/src/cpuinfo/SDL_cpuinfo.c: In function 'SDL_GetSystemRAM':
/?BUILDDIR?/libsdl2-2.0.1+dfsg1/src/cpuinfo/SDL_cpuinfo.c:632:35: error: 'HW_MEMSIZE' undeclared (first use in this function)
int mib[2] = {CTL_HW, HW_MEMSIZE};
^
/?BUILDDIR?/libsdl2-2.0.1+dfsg1/src/cpuinfo/SDL_cpuinfo.c:632:35: note: each undeclared identifier is reported only once for each function it appears in
make[2]: *** [build/SDL_cpuinfo.lo] Error 1
Tim McDaniel
On OSX, an SDL app forces a system shutdown to be cancelled. This happens because [SDLAppDelegate applicationShouldTerminate] returns NSTerminateCancel. A better approach is to subclass NSApplication and override terminate to do nothing except call SDL_SendQuit. In response to a system shutdown notification, this allows the normal SDL Quit event processing to occur, and if the app then terminates, system shutdown occurs normally. Please see the attached patch, based on SDL 2.0.1.
Tim McDaniel
On OSX, the app name in the menu bar is not localized. This can be fixed using the following implementation for GetApplicationName in SDL_cocoaevents.m:
static NSString *
GetApplicationName(void)
{
NSDictionary *dict;
NSString *appName = 0;
appName = [[NSBundle mainBundle] objectForInfoDictionaryKey:@"CFBundleDisplayName"];
if (!appName)
appName = [[NSBundle mainBundle] objectForInfoDictionaryKey:@"CFBundleName"];
if (![appName length])
appName = [[NSProcessInfo processInfo] processName];
return appName;
}
- Use the SDL_AUDIO_MASK_DATATYPE bit when selecting an implementation
where it matters. Previously two existing AUDIO_F32 cases had been
written, but were unreachable.
- Add AUDIO_F32 case for SDL_ConvertSurround_4.
- Fix incorrect pointer arithmetic causing the 2 to 6 channel
conversion for 4 byte audio formats to read and write beyond the end
of the buffer.
Leonardo
Event watchers are being executed on the inverse order they are added because they are added to the head of the SDL_event_watchers list.
Since watchers are allowed to change events before they are reported (they shouldn't, imo), this breaks code that rely on watcher execution order (such as distributed event handling).
An easy scenario to see this behaving weird to the user is if you add an event watcher to check mouse coordinates and check them again in your event loop. If you add the watcher after renderer's one (which always happens after you have initialized renderer), you get the same event but different coordinates.
The proposed patch adds the event watcher in the tail of the list, not in the beginning, and correctly fixes this problem.
Alex Szpakowski
Currently the UIKit/EAGL backend for SDL's OpenGL context creation API doesn't support OpenGL ES 3, despite iOS 7+ being capable (on devices with the necessary hardware.)
I have attached a patch to add support. It's also slightly more future-proof, so eventual OpenGL ES 4+ capability on iOS should hopefully work without requiring changes to SDL's UIKit/EAGL backend.
Alex Szpakowski
On my Mac OS X system (10.9.1), the SDL_MOUSEWHEEL event reports negative X values when my trackpad scrolls to the right, and positive X values when my trackpad scrolls to the left. This is backwards from what I'd expect, and I don't think it matches the Windows wheel events.
The vertical scroll values are what I'd expect though, and are consistent what gets reported on Windows (positive Y for scrolling up, negative Y for scrolling down.)
This is with "scroll direction: natural" disabled in the OS X trackpad settings (i.e. my scroll direction in non-SDL OS X programs matches what happens in Windows and Linux.)
I also tested with the horizontal scroll on a real mouse (Logitech G500 without custom drivers), and the horizontal scroll values in SDL are still flipped.
I "solved" the issue for myself by changing this line in the Cocoa_HandleMouseWheel function:
float x = [event deltaX];
to this:
float x = -[event deltaX];
I believe it should work fine with that change - I found something similar in another codebase while looking online for my issue - but I haven't tested on anything below Mac OS 10.8.
Besides being a little more simple to use than the earlier IOKit HID API, and
less likely to be deprecated, it also has the added benefit of working with
the Sony DualShock 4 controller in Bluetooth mode out of the box, whereas
the previous API has a bug that makes it report bad data for the
controller.
Cleaned up several other things in this code, having gone over every line of
it. The remaining deprecated calls are also gone.
Testing:
* Set the SDL_HINT_MOUSE_RELATIVE_MODE_WARP hint true, run testsprite2, press Ctrl-R to enter relative mode, alt tab away from the window, then click on the title bar of the window. Didn't get the mouse button release before, and we do now.
CR: Yahn + Alfred
Sandu Liviu Catalin
I'm unable to compile the latest SDL (directly from the repository) even though I disabled every DirectX option since I don't need DirectX.
I allways het these errors:
D:\DevLibs\SDL\src\render\direct3d\SDL_render_d3d.c:1897:1: error: unknown type name 'IDirect3DDevice9'
D:\DevLibs\SDL\src\render\direct3d\SDL_render_d3d.c:1898:25: error: unknown type name 'SDL_Renderer'
ny00
SDL_RenderClear clears a render target with the wrong color, if the opengles2 renderer driver is used and the target texture's format is SDL_PIXELFORMAT_ARGB8888.
The bug is *not* reproduced if SDL_PIXELFORMAT_ABGR8888 is used as the texture format (the first from the renderer's list).
It is further not reproduced using any of the following renderer drivers: opengl, opengles (apparently powered by Gallium3D), software.
Finally, the correct color can be drawn using SDL_RenderFillRect (instead of SDL_RenderClear).
A few details about the current setup:
- OS: Ubuntu 12.04 for x86_64
- GPU: GeForce GTX 460
- GPU driver version: 331.20-0ubuntu1~xedgers~precise1 (from the xorg-edgers PPA)
---
Seth Williams
Sam,
It appears that the clear just needs to take the render target format into consideration.
Seth.
Haneef Mubarak
AVX is the successor to SSE* and is fairly widely available. As such, it really ought to be detectable.
This functionality ought to be trivial to implement, and not having it means being forced to write an ugly workaround to check for AVX (so that normal SSE can be used if AVX is not available).
Here is an example on detecting AVX from SO (it actually shows ways to cehck for all of teh fancy instructions):
http://stackoverflow.com/questions/6121792/how-to-check-if-a-cpu-supports-the-sse3-instruction-set
The reasoning behind this change is that source2 in -tools mode has a single OpenGL context that is used with multiple different windows. Some of those windows are created outside the engine (i.e. with Qt) and therefore we need to use SDL_CreateWindowFrom() to get an SDL_Window for those. The requirement for sharing an OpenGL context across multiple different windows is that each window has the same pixel format. To facilitate this, I now set SDL_HINT_VIDEO_WINDOW_SHARE_PIXEL_FORMAT for the main window before calling SDL_CreateWindowFrom(). When I do this, SDL_CreateWindowFrom() will:
1. Set the pixel format of the returned window to the same pixel format as this SDL_Window passed in with the hint
2. The flag SDL_WINDOW_OPENGL will be set on the new window so it can be used for OpenGL rendering.
I only currently implemented this for Win32/WGL so implementing it for other platforms (i.e. X11) remains a TODO.
CR: SamL
Some pseudocode that shows how this is used in Source2:
HWND hExternalHwnd; // HWND that was established outside of SDL
// Create main window (happens inside platwindow.cpp)
SDL_Window *mainWindow = SDL_CreateWindow( , SDL_WINDOW_OPENGL .. );
// Create GL context, happens inside rendersystemgl
SDL_GLContext onlyContext = SDL_GL_CreateContext( mainWindow );
// Now I need to create another window from hEternalHwnd for my swap chain that will have the same pixel format as mainWindow, so set the hint
SDL_SetHint( SDL_HINT_VIDEO_WINDOW_SHARE_PIXEL_FORMAT, CFmtStr( %p, mainWindow) );
// Create the secondary window. This returned window will have SDL_WINDOW_OPENGL set and share a pixel format with mainWindow from the hint
SDL_Window *secondaryWindow = SDL_CreateWindowFrom( hExternalHwnd );
// To render to the main window:
SDL_GL_MakeCurrent( mainWindow, onlyContext );
// Do some rendering to main window
// To render to the secondary window:
SDL_GLMakeCurrent( secondaryWindow, onlyContext );
// Do some rendering to secondary window
This fixes weird behavior on Mac where our first responder reverts to the window
itself, rather than the SDLTranslatorResponder, after the window has lost focus
once. This causes Escape to call cancelOperation: on the NSWindow, which by
default removes our fullscreen-ness.
When someone has turned off SDL_TEXTINPUT we should probably set another initial
responder that handles the Escape behavior, so that SDL_TEXTINPUT doesn't change
fullscreen behavior (and possibly other behavior) like it does now.
Previously, we'd minimize fullscreen windows (if
SDL_HINT_VIDEO_MINIMIZE_ON_FOCUS_LOSS was set) during SDL_DestroyWindow if they
had keyboard focus, because we call SDL_SetKeyboardFocus(NULL) which yields a
OnWindowsFocusLost event.
Related to https://bugzilla.libsdl.org/show_bug.cgi?id=1840
SDL 2.x recently accepted patches to enable OpenGL ES 2 support via Google's ANGLE library. The thought is to try to eventually merge SDL/WinRT's OpenGL code with SDL-official's.
Leszek Godlewski
As described in the other thread
(http://lists.libsdl.org/pipermail/sdl-libsdl.org/2013-November/091997.html),
I've run into a case of SDL2 not recognizing a wireless Xbox 360
controller receiver properly on Debian Linux amd64 testing.
Apparently, the generated GUID is slightly different.
Device in question:
Bus 001 Device 015: ID 045e:0291 Microsoft Corp. Xbox 360 Wireless
Receiver for Windows
BurnSpamAddress
Steps to reproduce:
1. Grab the cursor with SDL_SetCursorGrab()
2. Alt-tab away from the window
3. Click on the titlebar of the window
This will cause the window to disappear underneath the taskbar!
This appears to be a general issue with ClipCursor() on windows, i.e. I am getting the same behavior if I call ClipCursor() directly.
It is caused by a feedback loop between the ClipCursor function and the modal resize/move event loop that handles mouse-based sizing on Windows.
Ghassan Al-Mashareqa
The SDL_ceil function is implemented incorrectly when HAVE_CEIL is not defined (HAVE_LIBC not defined).
The following code:
double val = SDL_ceil(2.3);
printf("%g", val);
prints "2.0", as STD_ceil is defined as:
double
SDL_ceil(double x)
{
#ifdef HAVE_CEIL
return ceil(x);
#else
return (double)(int)((x)+0.5);
#endif /* HAVE_CEIL */
}
This functions is used in the SDL_BuildAudioResampleCVT function of the audio subsystem (SDL_audiocvt.c), and causes a bug in that function.
Lets Android take care of which is the primary pointer (the one acting as the
mouse in SDL), reorganized the Java side code as well to make it easier to
understand.
If your application wants to have access to the full resolution even when the system has DPI scaling enabled, call SetProcessDPIAware() before calling SDL_Init()
e.g.
typedef BOOL (WINAPI *SetProcessDPIAware_t)(void);
HMODULE hMod = LoadLibrary("user32.dll");
if ( hMod ) {
SetProcessDPIAware_t pSetProcessDPIAware = GetProcAddress( hMod, "SetProcessDPIAware" );
if ( pSetProcessDPIAware ) {
pSetProcessDPIAware();
}
FreeLibrary( hMod );
}
To enable this, set the environment variable SDL_MOUSE_RELATIVE_MODE_WARP to "1"
When mouse relative mode is disabled, put the cursor back where the application expects it to be, instead of where it was when relative mode was enabled.
Based on the original port to Wayland by: Joel Teichroeb, Benjamin Franzke, Scott Moreau, et al.
Additional changes in this commit, done by me:
* Wayland uses the common EGL framework
* EGL can now create a desktop OpenGL context
* testgl2 loads GL functions dynamically, no need to link to libGL anymore
* Assorted fixes to the Wayland backend
Tested on the Weston Compositor (v1.0.5) that ships with Ubuntu 13.10,
running Weston under X. Tests ran: testrendercopyex (all backends), testgl2, testgles2,testintersections
pjz
SDL_ltoa(-2147483648,s,10) only returns "-" because there is a bug in the code:
if ( value < 0 ) {
*bufp++ = '-';
value = -value;
}
but -(-2147483648) is still -2147483648 (0x80000000) as signed int (or long), so the following loop doesn't run at all. Similar bug are also in SDL_lltoa.
BTW, there is no sanity check for radix.
SDL_GameControllerAddMappingsFromFile is now a convenience macro.
controllermap can now skip bindings by pressing space or clicking/touching the
screen.
Ryan C. Gordon
To keep the directory layout sane, we should probably move this one piece of source to the linux catch-all directory, instead of making it look like this is part of an SDL "input" subsystem.
norfanin
The MMX path in SDL_fillrect.c uses the SSE intrinsic _mm_stream_pi. The function or symbol provided by the compiler will not be present because the SSE header may not get included. The linker will complain about an undefined reference.
Since this is the only intrinsic used here (and someone forgot to create one for MOVQ), I think the MMX path can be removed completely. At least I don't see another way to move 64-bits from an MMX register to memory.
philhassey
On OS/X after calling SDL_SetWindowBordered right mouse clicks no longer register.
Steps to Reproduce:
1. Open a windowed window on OS/X. (With the border on.)
2. e.button.button will give values 1,2,3 depending on which mouse button I click.
3. Call SDL_SetWindowBordered to disable the border.
4. e.button.button will only give values 1,2. 3 (right mouse button) stops coming through.
Expected result:
I expect all mouse buttons to register.
Haiku uses most of the standard pthread API, with a few #ifdefs where we
still need to fallback onto the old BeOS APIs.
BeOS, however, does not support pthreads (or maybe doesn't support it well),
so I'm unplugging support for the platform with this changeset. Be Inc went
out of business in 2001.
Ivan Rubinson
As it turns out, it was impossible to render a texture flipped diagonally (both vertically and horizontally) with one RenderCopyEx call.
With help from #SDL @ freenode, we came up with a fix.
"Terminating app due to uncaught exception 'NSInternalInconsistencyException', reason: 'backgroundWindows not nil in enterFullScreenTransitionWithOptions:animated:activatingIt:'"
To reproduce this, run testsprite2, press Alt-Enter once, again while it's animating to fullscreen, and then again while it's animating out of fullscreen.
philhassey
Overview: While SDL_SetTextInputRect works perfectly to move my window out of the way of the virtual keyboard using SDL2/Android, on iOS this function has no effect.
Steps to Reproduce: Call SDL_SetTextInputRect with a rect near the bottom of the screen before calling SDL_StartTextInput.
Actual Results: The iOS virtual keyboard is displayed after calling SDL_StartTextInput, but the screen is not shifted to reveal the TextInputRect region.
Expected Results: The screen should be shifted to reveal the TextInputRect region (like with SDL2/Android.)
This patch implements SDL_SetTextInputRect for uikit/iOS.
It sets up notification handlers to respond to changes in the display of the keyboard. These handlers then change the frame of the view so it is moved out of the way of the keyboard as per SetTextInputRect.
This fixes behavior with the new Mac OS X fullscreen space code, as well as improve behavior on Linux desktops.
The default for normal fullscreen mode is still to minimize because we're likely doing a mode switch and don't want to stick around as a borderless window in the background.
I still haven't figured out why my application is being minimized when I try to raise, it but my previous workaround is causing issues.
For now the correct way to raise and/or restore the window is as follows:
if ( !(SDL_GetWindowFlags( window ) & SDL_WINDOW_MINIMIZED) )
{
SDL_RaiseWindow( window );
}
if ( SDL_GetWindowFlags( window ) & SDL_WINDOW_MINIMIZED )
{
SDL_RestoreWindow( window );
}
I will investigate the window state change rules more fully in the future.
CR: Alfred Reynolds
When we get a kCGEventTapDisabledByTimeout or
kCGEventTapDisabledByUserInput, the event tap would perform an invalid
memory access. void pointers are so fun.
This code only runs if you explicitly build with SDL_MAC_NO_SANDBOX.
Patrice Mandin
I encountered a problem trying to load a 8-bit paletted BMP file using SDL. This file was generated using GIMP 2.8. It has a big BITMAPINFOHEADER (0x6c bytes for biSize field), and thus the palette is incorrectly setup.
Lorenzo Pistone
this patch adds support for the horizontal wheel in Windows. It is shamelessly copied off the vertical wheel code, but I guess that that is a value added in consistency.
Thanks to Denis Bernard!
Also, changed the Android manifest so the app doesn't quit with orientation
changes, and made testgles.c exit properly on Android.
This bumps the build SDK level to 12 (up from 10). Runtime requirements remain
the same (at API level < 12 joystick support is disabled).
Also enables building SDL for armv7 and x86.
A port of the ANGLE library (OpenGL ES 2.0 for Direct3D) to WinRT, via https://github.com/stammen/angleproject, is used as a base.
To enable, clone 'angleproject' into the directory one above where SDL/WinRT is, open the file SDL/include/SDL_config_winrt.h, and uncomment the #defines that begin with 'SDL_VIDEO_OPENGL'. From there, apps can create an OpenGL capable SDL_Window via the flag, SDL_WINDOW_OPENGL, and an OpenGL ES 2 context via SDL_GL_CreateContext. The Direct3D 11.1 renderer cannot be used alongside SDL_WINDOW_OPENGL. Only Windows 8/8.1 is supported for now. Shaders may need to be precompiled, in some (all?) cases.
norfanin
The audio_enumerateAndNameAudioDevicesNegativeTests test in testautomation_audio.c reports a failure for SDL_GetAudioDeviceName when called on a driver that has only the default device. SDL_GetNumAudioDevices reports 1, but SDL_GetAudioDeviceName does not check if the index passed by the caller is in that range in this case. For positive numbers anyway.
This can be reproduced with the dummy driver on Windows and Linux.
Mai Lavelle
I've recently tried to create multiple windows and process key events for them, and found that key events weren't being generated for most of the windows. After some investigating I've observed the following effects. All but the most recently created window experience these effects...
- a focus lost event is generated immediately after the focus gained event, even tho window still has focus
- key events report window id 0 rather than the id of the window which has focus, SDL thinks no window has focus?
- giving focus to a non SDL window and then selecting an SDL window causes events to be generated as expected, but only until focus changes again
Focus change events are queued and delayed (200 ticks) before they are dispatched. The problem occurs when a focus out and focus in event are received on the same tick. When these delayed events are dispatched they will be sent in the order determined by the window list rather than the order in which they are received.
The focus out dispatch is implemented by calling SDL_SetKeyboardFocus(NULL). This will remove focus from any window, regardless of whether it is the one originally targeted by the X11 event.
Since SDL_SetKeyboardFocus() will always dispatch a focus lost event as needed, the easiest solution is simply to only call SDL_SetKeyboardFocus(NULL) when SDL_GetKeyboardFocus() matches the target window.
The first bug had mouse motion events not getting sent out on non-touch devices, if and when a mouse button wasn't pressed.
The second bug caused virtual mouse motion events to get sent out-of-order on touch devices: the motion event would get sent after the touch occurred, rather than before.
To enable the Debug Layer, set the hint, SDL_HINT_RENDER_DIRECT3D11_DEBUG to '1'.
The Debug Layer will be turned off by default, both in Release and Debug builds (of SDL).
Nitz
I was going through the SDL_IntersectRectAndLine function and wondered to see the ComputeOutCode function implementation.
The problem in this algo is, x and y axis are getting check with respect to 0, Which is wrong, it should be get checked with respect to rectangle x and y axis.
C.W. Betts
The recommended way of getting a file name that POSIX file APIs can open in OS X when using an NSString is -[NSString fileSystemRepresentation]. However, the current filesystem API in hg uses -[NSString UTF8String].
Andreas Ertelt
The problem in question is caused by changeset 7771 (http://hg.libsdl.org/SDL/rev/5486e579872e / https://bugzilla.libsdl.org/show_bug.cgi?id=2121)
The redefinition of __inline__ (introduced by the addition of begin_code.h:128's "|| __STRICT_ANSI__") results in mingw's gcc throwing multiple
warning: always_inline function might not be inlinable [-Wattributes]
as well as a whole bunch of redefinitions of mingw internals which break linking of projects including the SDL2 headers.
Sean McKean
I am running Ubuntu 12.04 (GL version 1.4 Mesa 8.0.4) , and on drawing a set of lines through the renderer through SDL_RenderDrawLines() (looped or not) or SDL_RenderDrawRect() I notice a pixel missing. For RenderDrawLines() it seems to be the second point in the sequence; for RenderDrawRect() it is the lower-right. This can be fixed by specifying SDL_RenderDrawPoint(s), but wouldn't it be easier to specify each pixel in a GL_POINTS glBegin/End loop in the OpenGL code, just to make sure?
I also ran the same program on Android; the rendering seemed to be correct, which uses glDrawArrays.
Kevin Wells
Overview:
SDL_RenderClear is only clearing part of a texture when it is the render target and a different size than the screen.
Steps to Reproduce:
1) This only occurs with the render driver set to direct3d, so: SDL_SetHint(SDL_HINT_RENDER_DRIVER,"direct3d")
Also, my window was 1280x720.
2) Create a texture for a render target with a resolution of 1024x1024:
texture=SDL_CreateTexture(main_window.renderer,SDL_PIXELFORMAT_RGBA8888,SDL_TEXTUREACCESS_TARGET,1024,1024);
SDL_SetTextureBlendMode(texture,SDL_BLENDMODE_BLEND);
3) Target the texture for rendering: SDL_SetRenderTarget(main_window.renderer,texture);
4) Set the draw color to whatever you want (problem occurs with both 0,0,0,0 and 0,0,0,255 among others) and then clear the render target:
SDL_SetRenderDrawColor(main_window.renderer,0,0,0,0);
SDL_RenderClear(main_window.renderer);
Actual Results:
Only about the top 3/4s of the texture gets cleared on calling SDL_RenderClear. The bottom 1/4 or so does not clear.
Expected Results:
Entire render target should be cleared.
J?nis R?cis
Brief history:
We recently ported a game from SDL 1.2 to SDL 2. While doing Windows testing, I soon discovered that the game exits without opening a window with my cross-compiled SDL2.dll, but works great with the SDL2.dll from the MinGW SDK on libsdl.org. It was as simple as swapping out the DLLs to make it work.
Running the game in Wine showed that the game actually does run, up until the call to SDL_CreateWindow, which fails and leads the game to print out an error:
Failure to create window (LoadLibrary("OPENGL32.DLL"): (null))
Which basically says that there was no error, but maybe that's a Wine quirk.
The error string originates in SDL_windowsopengl.c, in WIN_GL_LoadLibrary, which contains this piece of code:
wpath = WIN_UTF8ToString(path);
_this->gl_config.dll_handle = LoadLibrary(wpath);
SDL_free(wpath);
if (!_this->gl_config.dll_handle) {
char message[1024];
SDL_snprintf(message, SDL_arraysize(message), "LoadLibrary(\"%s\")",
path);
return WIN_SetError(message);
}
After some digging, I discovered the culprit: WIN_UTF8ToString returns NULL. Why? Because it calls iconv_open from an iconv.dll that does not support the UCS-2-INTERNAL encoding. Why does the official SDL2.dll work? Because it calls no external iconv functions at all.
It turns out that the Fedora MinGW infrastructure (from which I obtained the conventiently prebuilt iconv.dll) does not provide a DLL from libiconv, but instead provides a DLL from a minimal Windows library called win-iconv. Which knows a good bit, but doesn't know anything about UCS-2-INTERNAL:
http://code.google.com/p/win-iconv/source/browse/trunk/win_iconv.c#155
So there are two problems here:
1) The error message is clearly useless, because LoadLibrary is an innocent bystander. Instead wpath should probably checked for NULL, and a more appropriate error should be set. Ideally something that makes it clear than an external iconv is causing trouble.
2) SDL doomed itself at the ./configure step, by finding an existing iconv and happily using it without confirming support for the mandatory encodings required by SDL.
There are certainly a few easy ways out of the situation (although I didn't yet manage to figure out how to prevent ./configure from looking for external iconv), but this had me completely stomped for a good while, so I figured it's worth writing down if anything.
(Search also found this, which talks a little about using UTF-16LE instead of UCS-2-INTERNAL: https://bugzilla.libsdl.org/show_bug.cgi?id=2075)
Andreas Ertelt
SDL_dxjoystick.c is setting the classguid for device (dis)connect events to USB Devices in general:
dbh.dbcc_classguid = GUID_DEVINTERFACE_USB_DEVICE;
Wouldn't it make more sense to have it just subscribe to Hid device events? This would mean less meaningless events reaching the application.
Marcus von Appen
Revision eecbcfed77c9 of the SDL hg repo introduces the new
SDL_GetSystemRAM() function, which breaks the build on FreeBSD. Find
attached a patch, which unbreaks the build and also should (for most
cases) properly implement the sysctl support it.
David Ludwig
CarniBlood
Sam Lantinga
J?rgen P. Tjern?
Gabriel Jacobo
Ryan C. Gordon
James Legg
Edward Rudd
Brandon Schaefer
Philipp Wiesemann
dbrady
Pierre-Loup A. Griffais
Stefanos Apostolopoulos