| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
native handle via own ctor, reducing redundancy
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
and ctxOptions
- State-less operation during profile probing (mapGLVersions).
While probing the GLDrawable/GLProfile and GL instance
may not reflect the currently probed OpenGL profile.
Hence stateless operation by passing required information
is required for:
- GLDynamicLookupHelper must be fetched via
'major-version and contextOptions'.
- GLContextImpl.resetProcAddress(..)
- GLContextImpl.updateGLXProcAddressTable()
- GLContextImpl.setGLFunctionAvailability(..)
- ExtensionAvailabilityCache
TODO: Add replacement for GLProfile validation,
which is disabled right now.:
drawable.getGLProfile().verifyEquality(gl.getGLProfile())
The GLDrawable.GLProfile maybe less than GL's GLProfile
due to current context-version and options.
Hence we would need a 'GLProfile.bwCompatibleWith(GLProfile)'.
|
|
|
|
|
|
|
|
| |
On networking windowing systems (X11), we shall utilize the real
native default display connection.
On X11, this is X11Util.getNullDisplayName(),
for other non networking types, this is AbstractGraphicsDevice.DEFAULT_CONNECTION.
|
|
|
|
|
|
|
|
|
|
| |
AbstractGraphicsDevice's uniqueID instead of connection
AbstractGraphicsDevice's uniqueID supports multiple device types and units while including the connection.
This is required for proper EGLDrawableFactory's profile probing on different native platform devices.
Using common abstract implementation 'AImplementation'
for WindowsWGL's, X11GLX's and EGL's DrawableFactory.
|
|
|
|
|
|
|
|
| |
probeSurfacelessCtx
Since no known NVidia driver is know properly supporting surfaceless context,
i.e. current context w/o drawable attached, we set quirk NoSurfacelessCtx
for all NVidia drivers until a 'good one' is known!
|
|
|
|
| |
'-MapGLVersions' suffix.
|
|
|
|
|
| |
i.e. major >= 3 || hasGLVersionByString.compareTo(Version3_0) >= 0.
Otherwise the 'VersionNumber hasGLVersionByInt' was never used -> reduce possible GL_ERRORs.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
[GLX|WGL]_EXT_swap_control_tear extensions support
asynchronous buffer swaps, i.e. adaptive Vsync.
<https://www.opengl.org/wiki/SwapInterval_aka_vsync#Adaptive_Vsync>
<https://www.opengl.org/registry/specs/EXT/wgl_swap_control_tear.txt>
<https://www.opengl.org/registry/specs/EXT/glx_swap_control_tear.txt>
<http://keithp.com/blogs/async_flip/>
The extensions utilizes a negative interval value,
enabling late swaps to occur without synchronization to the video frame.
Hence '-1' has new semantics, previously it was the 'default value'
of 'untouched vsync interval'.
New default is:
- 0 for unrealized context
- 1 for realized context
+++
It requires [GLX|WGL]_EXT_swap_control,
hence we shall ensure to use use this extension
in the implementation of GLContext.setSwapInterval(..).
+++
Mesa3D seems to support GLX_SGI_swap_control only.
+++
Implemented on Windows and X11.
+++
On GNU/Linux using NVidia driver w/ my setup(*), sadly the query
GLX.glXQueryDrawable(displayHandle, drawable.getHandle(), GLX.GLX_LATE_SWAPS_TEAR_EXT, val);
always returns zero here, indicating async vsync is not supported.
(Queried the attribute for every frame in windowed or fullscreen mode)
Fullscreen
(*)
- Debian 8
- Kernel 3.16
- KDE/Kwin
- GL Version 4.5 (Core profile, arb, compat[ES2, ES3, ES31], FBO, hardware) - 4.5.0 NVIDIA 355.06 [GL 4.5.0, vendor 355.6.0 (NVIDIA 355.06)]
- GL_RENDERER GeForce GTX 660/PCIe/SSE2
- Samsung U28D590 (DFP-4): Internal DisplayPort
+++
|
| |
|
|\ |
|
| |
| |
| |
| | |
Signed-off-by: Xerxes Rånby <[email protected]>
|
| |
| |
| |
| | |
Signed-off-by: Xerxes Rånby <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
(Fullscreen on Bottom not working though)
- Terminology
- ALWAYSONBOTTOM -> isOnBottom
- ALWAYSONTOP -> isOnTop
- Better support for 'isOnBottom' on Windows
- Refine 'SetWindowPos(..)' calls to handle isOn[Top|Bottom]
- Refine 'ShowWindow(..)' calls to handle isOnBottom
- Intercept WM_WINDOWPOSCHANGING to enforce isOn[Top|Bottom],
i.e. change params (hwndInsertAfter and flags).
- Fullscreen on Bottom not working
- While we change all flags appropriately (NOACTIVATE, ..)
and don't issue CDS_FULLSCREEN is on bottom,
Windows seems to enforce the window to be on top
if it is sized to the screen maximum.
|
| |
| |
| |
| |
| |
| |
| |
| | |
- Fix STATE_MASK_CREATENATIVE
- Simplify resetStateMask()
- Windows WindowDriver: Remove STATE_MASK_STICKY from supported states
- TestGearsNEWT: Use NEWTDemoListener
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
getSupportedReconfigMaskImpl()
- See getSupportedStateMask() for semantics
- getSupportedReconfigMaskImpl() result now used for isReconfigureMaskSupported()
- getSupportedReconfigMaskImpl() implementations:
- Full featured: X11, Windows and OSX
- X11: Use WindowManager Atoms for certain features (dynamic)
- Others: Use fixed features.
- TODO: Consider avoiding actions if not supported.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Unresolved strncpy_s (MSVCRT) on WinXP,
as shown w/ dependency walker (red module, red unresolved line).
Mapped: _tcsncpy_s -> strncpy_s (!UNICODE).
On WinXP MSVCRT has no strncpy_s.
_tcsncpy_s(sOut, sOutLen, s, len)
-> bound-check + _tcsncpy(sOut, s, len)
|
|\ \ |
|
| | | |
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| | |
- STATE_BIT_FULLSCREEN_SPAN is private and used for reconfigure,
hence STATE_BIT_COUNT_RECONFIG is needed.
- STATE_BIT_FULLSCREEN_SPAN is added at the end of public state bits
- PSTATE_BIT_MINMAXSIZE_SET is unused.
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
test.
- ImageIOUtil -> ImageType + ImageType.Util
- ImageType.Util.getFileSuffix(..):
- Fix byte type conversion, i.e. 'b == (byte)0x89',
cast is required to avoid byte -> int conversion.
Note: signed byte -128 - +128
- Parse in O(1), i.e. lexicographical parsing
- FIXME: We seem to have at least three type collisions, validate!
- ImageType:
- Complete T_* w/ API doc -> FIXME/TODO missing type references!
- ImageType instancing via InputStream or manual type definition.
- TextureData
- Contains optional source ImageType
- TextureProvider:
- Deprecate newTextureData(..) variants other than InputStream
simplifying TextureIO.
- TextureProvider.SupportsImageTypes:
- Added interface, allowing mapping ImageType -> provider
- Tested standalone ImageType (TestImageTypeNEWT) and
via TextureIO (TestTextureIONEWT) utilizing list of all
test data (ImageTstFiles), i.e. PNG, JPG, TGA and DDS.
|
|/ |
|
| |
|
| |
|
|
|
|
| |
@since 2.3.2 tag
|
|\ |
|
| | |
|
| | |
|
|\ \ |
|
| |/
| |
| |
| | |
empty data
|
|/ |
|
|
|
|
|
|
|
| |
- to avoid resizing when toggling resizable (change of window border)
we maintain 'RECT insets' in our WindowUserData struct
and fix the client -> top position late in NewtWindow_setVisiblePosSize(..)
after any style change.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
WindowImpl
- remove updateMinMaxSize(..) - unused info
- fix appendStateToString: show all maximized state changes if reconfig
- add sizePosMaxInsetsChanged(..) and sendMouseEventRequestFocus(..)
accumulating multiple callbacks from impl.
- add: maximizedChanged(..) notification from native impl.
- refine manual maximized mode
used for OSX and Windows (single extent)
- reconfigMaximizedManual(..)
- resetMaximizedManual(..)
X11 WindowDriver:
- Update maximized at xreconfig, read from _NET_WM_STATE
- Use less Java callbacks from JNI
Windows WindowDriver:
- Use native maximized, if HORZ && VERT,
otherwise use manual maximized for single extent.
- Invisible of top-window -> MINIMIZED/ICONIFY
showing the app in task-bar.
|
|
|
|
|
|
| |
rely solely on insetsChanged(..) (event driven)
Only affected WindowDriver is AWT, which now updates the insets when setVisible(true).
|
|
|
|
|
|
| |
If not waiting for resize on main-thread,
corruption may happen at continued rendering.
This has been experienced w/ maximize-horizontally (Bug 1188).
|
|
|
|
| |
reconfig DEBUG
|
| |
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Bug 1178 cc10: We no longer throw an ExceptionInInitializerError
when X11 is not available.
Fix 1178 cc10: We need to use an X11 resource in the constructor
in order to throw an ExceptionInInitializerError if X11 is not available.
We can resolve this by query for the X11 display and screen inside the constructor.
Signed-off-by: Xerxes Rånby <[email protected]>
|
| |
| |
| |
| | |
states, dump current and new states upfront.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
child windows / Fix [Un}Resizable style
- Fix regression on OSX setVisible: in-visibility never reached on child windows
- Fix 'typo' while porting to bit-mask, i.e. '!=' -> '=='.
- Fix [Un}Resizable style
- Use NSResizableWindowMask only for !Undecorated and if STATE_MASK_RESIZABLE is set.
|
| | |
|
|/
|
|
|
|
|
|
|
| |
child windows
It has been experienced that UnmapNotify is not sent for child windows when using IconicState!
Hence the visible:=false event never reaches the Window, causing an error.
This patch only uses IconicState for top-level windows and if requested.
|
|\ |
|
| |
| |
| |
| | |
report 1042
|
| |
| |
| |
| | |
the bug report 1042
|