Commit graph

5491 commits

Author SHA1 Message Date
Sam Lantinga
5d20dc56bf Get the correct screen bounds from xinerama 2012-09-28 14:22:18 -07:00
Sam Lantinga
403f5ab53d Print the bounds of each display when showing mode information 2012-09-28 14:21:15 -07:00
Sam Lantinga
2c196568a8 We need to queue the focus in/out changes because they may occur during video mode changes and we can respond to them by triggering more mode changes. 2012-09-28 14:01:55 -07:00
Sam Lantinga
b40f346c7e Don't warn about the window size changing if the window is resizable 2012-09-28 14:00:43 -07:00
Sam Lantinga
49cba901a5 Fixed detection of display bounds after a mode switch when Xinerama is enabled. 2012-09-28 11:51:16 -07:00
Ryan C. Gordon
3861a6c17c Patched to compile on Visual Studio. 2012-09-28 14:17:30 -04:00
Ryan C. Gordon
f75687d3e7 Add a safety net X11 error handler, to reset the vidmodes before crashing.
--HG--
extra : rebase_source : 750eb796a465f53d50d47b297ccd3f17aeda6f9e
2012-09-28 14:12:45 -04:00
Sam Lantinga
a6fdbb6213 Theoretically fixed the position of the child window, though it's always ending up at the upper left corner for some reason. 2012-09-28 10:54:26 -07:00
Sam Lantinga
26efe14fd6 A few more tweaks for the legacy fullscreen support 2012-09-28 10:24:28 -07:00
Sam Lantinga
9bea482166 Fixed bug 1579 - Creating a texture with unsupported format may cause double-destruction
Alexander Hirsch 2012-08-25 20:01:29 PDT

When creating a SDL_Texture with unsupported format (I'll now refer to it as
texture A), SDL_CreateTexture will call SDL_CreateTexture again with
GetClosestSupportedFormat to set texture->native (which I will now refer to as
texture B).
This causes texture B to be put before A in renderer->textures.

If texture A is explicitly destroyed, everything is fine. Otherwise, upon
SDL_DestroyRenderer, the loop will first encounter texture B, destroy it, then
texture A, destroy that which will want to destroy texture->native and since it
is already destroyed set an error.

The solution could be as simple as swapping texture A with B after
texture->native gets set in SDL_CreateTextures.
2012-09-28 04:09:06 -07:00
Sam Lantinga
360f769275 Fixed bug 1605 - SDL_DestroyWindow causes erroneous error message
You can't pass a NULL window into SDL_GL_MakeCurrent()
2012-09-28 04:03:06 -07:00
Sam Lantinga
29b5947d0c Clear the queue active flag when we shutdown, so we don't have a mutex error trying to read the queue. 2012-09-28 03:54:39 -07:00
Sam Lantinga
c328979f60 Fixed bug 1584 - Improved glError checks in the opengl renderer
Martin Gerhardy 2012-08-27 02:42:25 PDT

I've extended the gl error checks.

This is needed because on my system there are errors in the renderer that are
hard to find.

Also glError can return multiple errors. Even if SDL_SetError would only
contain the last one of course, the SDL log facilities are able to get the
output for each error.
2012-09-28 03:49:27 -07:00
Sam Lantinga
400b953d29 Made it clear that locking a streaming texture is a write-only operation. 2012-09-28 03:29:36 -07:00
Sam Lantinga
049837bbfc Result of autogen.sh run with previous commit 2012-09-28 03:22:26 -07:00
Sam Lantinga
584f8bb2ce Fixed bug 1587 - "aclocal" fails to generate a sufficient "aclocal.m4"
Cecil Curry 2012-08-27 16:55:12 PDT

Allow "aclocal" to find the "AC_CHECK_DEFINE" macro.

On running:

   rm aclocal.m4
   libtoolize --install --copy --force
   aclocal -I acinclude
   autoconf -I acinclude

The former three commands succeed, but "autoconf" fails with:

   configure.in:198: error: possibly undefined macro: AC_CHECK_DEFINE
         If this token and others are legitimate, please use m4_pattern_allow.
         See the Autoconf documentation.

"aclocal" fails to find AC_CHECK_DEFINE and hence add such macro to
"aclocal.m4". Here is why:

* "acinclude/ac_check_define.m4" defines AC_CHECK_DEFINE via define() rather
than AC_DEFUN().
* "aclocal" fails to find macros defined via define().
2012-09-28 03:18:18 -07:00
Sam Lantinga
8fef68eb90 Fixed memory leak in an error case 2012-09-28 02:43:13 -07:00
Sam Lantinga
a3e3ca708c Fixed bug 1496 - SDLView not created for a Cocoa window, resulting in missing right mouse clicks
When you create an nswindow it automatically gets a default view, so we'll just create an SDLView if we're creating the window.
2012-09-28 02:33:42 -07:00
Sam Lantinga
43674fabab We don't always get the focus change events. Argh! 2012-09-28 02:15:44 -07:00
Sam Lantinga
e6a0c63fb7 Simplified the NetWM code a little bit, and it now operates on X11 Window 2012-09-28 01:51:51 -07:00
Sam Lantinga
32c3816470 Don't set the hidden netwm state, that would mean we don't want to be shown, which is bad. 2012-09-28 01:36:26 -07:00
Sam Lantinga
b9083adeb5 Pass the flags in, not the window pointer! 2012-09-28 01:15:48 -07:00
Sam Lantinga
71450f7c85 Don't base our resizable state on what the window manager decided, instead we want to base it on the size hints which we set. 2012-09-28 01:11:30 -07:00
Sam Lantinga
6380cc6759 Only map/unmap if the window isn't already in the correct state. 2012-09-28 00:57:47 -07:00
Sam Lantinga
978b87dccb Fixed bug where desktop resolution was never restored if Xinerama was enabled. 2012-09-28 00:40:47 -07:00
Sam Lantinga
c5d1dc43c3 Compositing window managers can show and hide windows without ever affecting the mapped state. However they do send NetWM protocol messages to indicate this is happening.
Also refactored the netwm state code so it's consistent between the places that use it.
2012-09-27 23:55:38 -07:00
Sam Lantinga
7e30790b41 Hopefully improved legacy fullscreen handling (it works now on my Ubuntu Unity system) 2012-09-27 17:17:33 -07:00
Sam Lantinga
535b566931 Removed unused variables 2012-09-27 14:38:56 -07:00
Sam Lantinga
f380ecb137 Removed executable bit from source files 2012-09-27 14:35:28 -07:00
Sam Lantinga
5390cf0ace When we're reparented we'll get both unmapnotify and mapnotify. Eat both so the unmap notify doesn't confuse the normal event processing. 2012-09-27 13:52:57 -07:00
Sam Lantinga
fb1d522bc3 Show events from all windows (for the legacy fullscreen case) 2012-09-27 13:52:07 -07:00
Sam Lantinga
2c07f9f35b Fixed fullscreen origin for multi-head displays 2012-09-27 13:23:34 -07:00
Sam Lantinga
1401f159d0 Added debug define (commented out) 2012-09-27 13:22:34 -07:00
Sam Lantinga
c95f18c57a Use the legacy fullscreen code if the window manager doesn't support the NetWM protocol. 2012-09-27 11:39:31 -07:00
Sam Lantinga
0582dc6ce3 Turned the X11 mode extension environment variables into hints so they can be more easily set from applications. 2012-09-27 11:29:13 -07:00
Sam Lantinga
86e52a87bf Print a message if the window size was changed from the requested size. 2012-09-27 11:13:30 -07:00
Sam Lantinga
8bbb66b684 Removed unused functions 2012-09-27 11:16:31 -07:00
Gabriel Jacobo
d36e0cf8b9 Don't block Android event loop until all relevant events reach the app. 2012-09-27 15:00:34 -03:00
Sam Lantinga
7c835f8560 Fixed bug where Xinerama was treated as being available even if it wasn't.
Also added additional logging for X11 video modes
2012-09-27 10:41:16 -07:00
Ryan C. Gordon
6d86238afc X11: Attempt to go fullscreen the way SDL 1.2 did it.
Ideally this code is never used, but as a legacy fallback, it could be useful.

--HG--
extra : rebase_source : aea45507c3b6388d929b6074351fb59c2dab2097
2012-09-27 03:36:13 -04:00
Ryan C. Gordon
170089c6aa On Compiz (etc?), _NET_WM_ACTION_FULLSCREEN needs the window to be resizable.
Thanks to Edward Rudd for the patch!

--HG--
extra : rebase_source : 2039db129b3a6801e0001c6b8f508f5b0d27269e
2012-09-27 00:53:37 -04:00
Sam Lantinga
9119439672 Fixed bug 1559 - SDL does not provide scancodes for all keys available on Android.
Philipp Wiesemann 2012-07-28 10:03:47 PDT

the current version of SDL 2.0 (HG) does not provide scancodes for all keys
available on Android.

Missing are all new keys added since Android API level 11.

Also it would be possible to map SDL_SCANCODE_WWW, SDL_SCANCODE_MAIL and
SDL_SCANCODE_SLASH.

Android's KEYCODE_DEL is mapped to SDL_SCANCODE_DELETE but according to
Android's documentation [1] it should be like SDL_SCANCODE_BACKSPACE.

I attached a patch with these corrections.
2012-09-26 20:28:58 -07:00
Sam Lantinga
ef6fa58028 Fixed bug 1573 - SDL does not support system clipboard on Android.
Philipp Wiesemann 2012-08-18 14:09:47 PDT

there is currently no way in SDL to interact with the system clipboard on
Android.

I attached a patch which tries to implement the three clipboard functions for
Android. It does not add the CLIPBOARDUPDATE event because this seems to
require Android API 11 or polling.
2012-09-26 20:14:37 -07:00
Gabriel Jacobo
255033689c Make SDL_atomic.h depend on the void pointer size determined at compile time
rather than at configure time. As OS X supports lipo'ed versions of SDL for i386
and x86_64, a single set of headers that work for both architectures is desired.
2012-09-26 15:57:28 -03:00
Edward Rudd
df2acac897 allocate *Hint structures per Xlib docs 2012-09-26 14:08:46 -04:00
Sam Lantinga
b603b9a9f9 The beeping stopped.
And, other people may need to catch key events using their own delegate.
2012-09-25 20:58:23 -07:00
Sam Lantinga
9f2bbcef0c Merged libtool fixes 2012-09-25 20:49:43 -07:00
Sam Lantinga
78d497098e This isn't necessary anymore, with the latest autotools update. 2012-09-25 20:49:29 -07:00
Sam Lantinga
144d68b081 The SDK path is no longer needed (and no longer correct) 2012-09-25 20:48:26 -07:00
Sam Lantinga
d2b7ba63d1 NetBSD patch to use AUDIO_GETBUFINFO when available (contributed by jmcneill) 2012-09-25 20:47:38 -07:00