| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
glMap*Buffer*; GLX/WGL/CgGL: All runtime dynamic; Misc ..
TODO: Compile and test on MacOSX ..
Fix:
=====
Multithreading/Locking:
See jogl/doc/Implementation/MultiThreading.txt
- Locking layer is not platform agnostic, ie GLContextImpl, GLDrawableImpl, ..
and NEWT: Window/Display
- No more use of JAWT global lock necessary, removed.
- No need for X11 Display lock, on the contrary,
this made the NV driver hang.
- Use common window/surface lock
- All NativeWindow surfaceLock's are recursive now
glMapBuffer: If size is 0, don't do cont with the native call.
glMapBufferRange: Fix capacity.
glNamedBufferDataEXT: Track the size.
glMapNamedBufferEXT: Manual impl. - use the tracked size
glXGetVisualFromFBConfig, glXChooseFBConfig, glXChooseVisual: Instead of
ignoring and implement a renamed version (*Copied), we just use ManualImplementation
for the proper copy-result code.
DesktopGLDynamicLookupHelper: Initialize _hasGLBinding* attributes
in the determing loadGLJNILibrary() method, which is called by super().
Otherwise static init will overwrite them after the super() call.
X11GLXDrawableFactory: Don't release anything at shutdown (removed sharedContext.destroy()),
since this caused a freeze/SEGV sometimes.
Fixed NEWT's reparentWindow() functionality incl NewtCanvasAWT usage.
- Native: if not visible, don't focus, etc
- NewtCanvasAWT: Use the container size to start with
- Run the command on the EDT
Using GlueGen's new DynamicLibraryBundle utility:
- X11, Windows and MacOSX OpenGL adapted to DynamicLibraryBundleInfo.
- X11GLXDynamicLookupHelper -> X11GLXDynamicLibraryBundleInfo
- Remove all path from lib names.
- GL order: libGL.so.1, libGL.so, GL
- shallLinkGlobal: true -> to server some 'old' DRI systems
-> http://dri.sourceforge.net/doc/DRIuserguide.html
- shallLookupGlobal: false
- Try both : glXGetProcAddressARB and glXGetProcAddress
- Using bootstrap: GLX.glXGetProcAddress(long glxGetProcAddressHandle, String glFuncName)
Found the issue with LIBGL_DRIVERS_PATH, ie if not set
no valid GL instance can be found (ie ATI fglrx/DRI).
This may happen if using a differen user than the desktop user
for whom the env var is set within some /etc/X11/Xsession.d/ script.
Enhancements:
=============
GLBufferSizeTracker: Use IntIntHashMap and add DirectState size tracking.
GLBufferStateTracker: Use IntIntHashMap.
GLStateTracker: Use IntIntHashMap.
GLDynamicLookupHelper: More generic (global loading/lookup and GetProcAddress function name list),
remove redundant code.
FIXME:
MacOSXCGLDynamicLookupHelper:
- Not tested
- Not using NSImage lookup anymore as recommended by OSX API Doc,
so dlsym is used always (to be tested)
WindowsWGLDynamicLookupHelper:
- Not tested
GLX/WGL/CgGL is all runtime-dynamic as now, ie loaded and looked-up at runtime,
no compile time dependencies to GL anymore, nor a need to specify CgGL.
Split up WGL in GDI and WGL, to allow proper dynamic runtime linkage of OpenGL32
while using static binding to GDI32
NEWT events generated by native code are enqueued and not send directly.
This should ease locking mechanisms .. if any are necessary.
NEWT: More platform specific code moved to *Impl method,
simplifying the generic code of the superclass and impl protocol.
Cleanup:
=========
Replace all InternalBufferUtil's with com.jogamp.common.nio.Buffers
Removed all InternalBufferUtil's from repository
Removed GLContextImpl notion of 'optimized' surface locking,
where the surface gets unlocked during makeCurrent/release.
This just makes no sense and would impact multithreading in a horrible way.
|
| |
|
|
|
|
|
|
|
| |
- Using Display.getEDTUtil() and Display.runCreateAndDestroyOnEDT()
to determine the NEWT EDT behavior, which may be specialized by the implementation.
- AWTWrapper and Newt/AWT Parenting deadlock fix.
- Misc fixes in test cases
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Misc.
Due to incapabilities of the previous AWT/NEWT reparenting
the implementation and spec had to be changed to support this feature.
See the first 2 comments below.
- Tested on GNU/Linux (OK), Windows (a few bugs left)
-
TODO:
- Clarify the size/layout issue, ie who is responsible etc
In the test, incl AWT/NEWT, we set the size on the GLWindow
and ie pack the AWT Frame.
- Fix remaining [Windows] bugs ..
- Fix/Implement MacOSX port ..
Fix AWT/NEWT reparenting:
===========================
- Now NewtFactory's createWindow() method for parenting handles NativeWindow only
and is no more responsible for creating a child window upon an AWT Component.
See class com.jogamp.newt.awt.NewtCanvasAWT for NEWT/AWT parenting.
- New com.jogamp.newt.awt.NewtCanvasAWT, responsible for handling
AWT's reparent events via addNotify/removeNotify.
Reparenting is implemented via the new NEWT Window's reparentWindow() method.
Also sets the background erase to false, if supported.
- Fix zero size semantics in Window (setSize/setVisible)
Since a zero size window is not supported by many compoenent (Windowing system, OpenGL, ..)
we use the visibility methodology to not show a 0x0 window. See Javadoc.
AWT components may start with zero size.
- New NEWT Window: reparentWindow(NativeWindow newParent, Screen newScreen)
Allowing to change the parent of a window. Similar with the fullscreen toggle,
but without size/position change.
Native reparenting allows to keep alive the native
window while changing the container, hence it is preferred to a destroy/create cycle.
To benefit from the native reparenting, a NEWT implementation has to implement
'protected boolean reparentWindowImpl(long newWindowHandle)'
and return true, otherwise reparenting will be 'emulated' via
the expensive destroy/create cycle.
- NEWT's Window references all of it's children, if any
- NEWT's Window propagates setVisible/destroy actions to it's children.
- Fix NEWT's destroy() semantics.
A call of destroy() or destroy(false) shall only result in the destruction of the
native window (handle) nothing more. A subsequent setVisible(true) shall allow
the complete recreation of the Window into a usable state.
A call of destroy(true) destroys all resources the Window holds,
may include Screen/Display and OpenGL resources in case of GLWindow.
This is necessary to allow proper reparenting, where a native window may become
destroyed, but should be recreated via setVisible(true) later on.
- Fix NEWT set[Size|Position|Fullscreen|Visible] synchronization.
Use a recursive lock instead of the Window instance, otherwise arbitrary Window access
via AWT's EDT, NEWT's EDT or other threads can block.
Also removed a use pattern like:
key.lock()
try {
EDT.invoke(action());
} finally {
key.unlock();
}
Where action() itself uses the same lock object (here key), the result is a deadlock.
NativeWindow Changes:
======================
- We can use XInitThreads() now (concurrent threading support)
in combination with AWT.
Might have been some async in our NEWT locking in regards to AWT (sync()),
and the X11 Display changes made in c787f50d77e2491eb0d8201d534a6fa4885a929e.
- NativeWindow's window handle is _not_ transient like surface handle,
fixed documentation.
JOGL Changes:
=============
- New 'isRealized()' method in GLDrawable.
-
Misc Fixes
============
- Fix NEWT set[Size|Position|Fullscreen|Visible] duplicate code
Due to pure abstract signatures, the set[Size|Position|Fullscreen|Visible]
implementations of X11, OSX, .. contained duplicate code and state handling (size, pos, ..).
These are now decoupled, ie generic set[Size|Position|Fullscreen|Visible] implementations
calling simple set[Size|Position|Fullscreen|Visible]Impl implementations.
- Fix NEWT: Renamed setAutoDrawableClient(boolean) to setHandleDestroyNotify(boolean)
The semantic of
setAutoDrawableClient(boolean) defaults to false
was too complicated and specific, hence changed to
setHandleDestroyNotify(boolean) defaults to true
since its more clear and the name refers the window itself..
- Fix NEWT: Removed GLWindow's unused global window list
- Fix NEWT: Remove Window's unused event mask
- Rename com.jogamp.newt.impl.awt.AWTNewtFactory -> com.jogamp.newt.awt.NewtFactoryAWT
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Use proper (X11) Display names to avoid multiple Display instances.
The problem was in case of 'wrapping' another X11 Display,
a previous Display instance used 'nil' to reflect the null default Display
but the wrapped instance (using a Display handle) the proper Display name.
Now all (X11) Display's are using the proper Display name instead of a dummy 'nil' name.
- Fix: NEWT null Display name is validated upfront, instead of changing it later
- Fix: Nativewindow's X11Util gathers the systems NULL Display name and offers
a validation method
- Fix: NEWT X11 Display validates the NULL Display name properly
|
| |
|
|
|
|
|
|
| |
- Restructure: JAWT gluegen, use common jawt-common.cfg
- Fix: Use proper capacity for GetDrawingSurface and GetDrawingSurfaceInfo
- Fix: JAWTWindow maintains AWT bounds
- Fix: JAWTWindow locking/unlocking decoupled with abstract implementation,
which allows clear code and simpler unlock code.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
+++++
Changed NEWT NativeWindow creation contract
Original contract:
(1) A native window was created as soon as possible,
ie when NEWTFactory.createWindow(..) was called - if possible.
(2) A valid native window has to be created at least after setVisible(true)
has been called.
Problems:
Not all implementation are able to create the native window that early,
but at setVisible(true) only (e.g: KD and EGL).
Due to window parenting especially the new AWT/NEWT parenting,
the native window can only be created in case the parent native window is valid.
New contract:
(1) A native window is created at setVisible(true),
if it's a top level window or the native parent window is valid.
(2) A valid native window may not be created after setVisible(true)
has been called.
Subsequent setVisible(true) calls shall be made in case
the creation has not been done yet.
This is demonstrated in GLWindow.display() for example.
The new contract implements a lazy native window creation.
+++++
AWT/NEWT Parenting
- HierarchyListener and ComponentListener ensure that the NEWT child window
will be setVisible according to the AWT parent window.
- Lazy native window creation relaxes requirements to the parent window's state.
- Attachment of the child window and setVisible() may be called
after NEWT child window creation.
- GLWindow supports NEWT child window creation directly
The test case TestParenting01AWT.java reflect this new contract
and demonstrates more simplified and more flexible use cases.
+++++
NEWT Fixes:
- All NEWT implementation's native code method names end with 0.
- GLWindow: don't issue the actual 'init'/'display' call to GLEventListeners
in case the window is not visible.
- NEWT setSize/setPosition:
if native-window
call native-window action and let the attributes being set by
the feedback call, which issues more action, ie RESIZE.
else
set the attributes directly, no feedback call/action is necessary.
+++++
X11 Fixes:
- X11GLContext MakeContextCurrent:
Use MakeCurrent in case write and read drawable are equal,
otherwise SEGV happens on ATI with heavy multithreading involved!
Even XLockDisplay and XSync didn't help here ..
- X11GLXDrawableFactory shared resource:
Removed the resource holder thread to simplify code,
hence proper release is no more desired and it could become a cause for deadlock.
- Moved XInitThreads() from NEWT X11Window -> NativeWindow X11Util,
since NativeWindow is loaded first (essential for XInitThreads())
and it is the more basic lib.
- Made call to XInitThreads() conditional, ie it's spared if
AWT could be used - which causes SEGV .. (AWT bug). See X11Util.java
+++++
JOGL Fixes:
- GLProfile.isAWTAvailable() -> NativeWindowFactory.isAWTAvailable()
- GLProfile.isAWTJOGLAvailable() -> GLProfile.isAWTAvailable()
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- X11GLXDrawableFactory:
- Move shared resource creation/destruction into it's own thread
- Remove the ATI hack (no XDisplay closing) for every Display,
this is only necessary for the shared XDisplay and in case of AWT.
- Newt
- Display: Only pumpMessages if device is ready.
- X11Display: Verify handle not null at DispatchMessage.
- Common recursive ToolkitLock implementation, from
src/nativewindow/classes/com/jogamp/nativewindow/impl/LockingNativeWindowFactory.java and
src/newt/classes/com/jogamp/newt/Window.java,
-> com.jogamp.nativewindow.impl.RecursiveToolkitLock
- Unique XLockDisplay/XUnlockDisplay call via X11Util to simplify debugging.
X11Util: Added debug code for XLockDisplay/XUnlockDisplay.
Added fast LongObjectHashMap
Added static lib loading and initialization.
Removed active and passive list, as well as unused methods,
to easy maintenance. Possible since the only 'uncloseable' Display
might be the shareable one.
- X11Lib: Added static initialization via X11Util
Test:
junit/com/jogamp/test/junit/jogl/demos/gl2/gears/TestGears*
- Add WindowListener for quit ..
|
| |
|
|
|
|
| |
- added lock(); try{foo();}finally{unlock();} where missing
- made sure lock is called outside the try block
- fixed nesting in situations with two independent locks
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- All available OpenGL versions (native/platform) are verified at GLProfile
initialization and can be queried ..
A mapping of major,compat -> major,minor,options is created.
- Removal of temp context creation, when creating a context.
This was necessary to query general availability of ARB_create_context.
Due to the shared context of X11GLXDrawableFactory and WindowsWGLDrawableFactory,
this is no more necessary.
Due to the version mapping, the ARB_create_context paramters are known.
- NativeWindow X11Lib: Added X11ErrorHandler, throwing a RuntimeException.
Necessary to catch BadMatch .. etc X11 errors, eg for glXCreateContextAttribsARB
Hence all X11 calls are covered now.
- X11DummyGLXDrawable needs to use an own Window, otherwise GLn n>2 fails
- Flattening the desktop GL* implementation,
all use GL4bcImpl, which reduces the footprint dramatically.
- GL*Impl.isGL*() (desktop) utilizes the GLContext.isGL*(),
hence the results reflect the actual native context version.
- GLContextImpl makeCurrent/create: Added workflow documentation,
clarified code, defined abstract methods to have a protocol.
- Removed moved files (from here to gluegen),
see gluegen a01cb3d59715a41153380f1977ec75263b762dc6
- NativeLibLoader -> <TYPE>JNILibLoader
- Fixed Exception Handling (as in gluegen bce53b52c8638729750c4286dbc04cb14329fd34),
ie removed empty catch Throwable ..
- GLContext.setSwapInterval(): Nop in offscreen case, otherwise X11IOError (NVIDIA Bug)
Test:
Tests
- Junit
- demos.gears.Gears
- demos.jrefract.JRefract
Platforms
- Linux 64/32 ATI/NVidia
- MacOsX
- Windows (virtualbox 3.1.6, offscreen failed)
TODO/BUGS:
- FIXME ATI GLn n>2 with AWT, can't make context current, works well on NVIDIA though
- FIXME GL3GL4: Due to GL3 and GL4 implementation bugs, we still choose GL2 first, if available!
- Add GL 3.3 to GL3/gl3ext.h
- Add GL 4.0 to GL3/gl3ext.h and fix the GL3/GL4 seperation
- Rename jogl.gl2.jar -> jogl.gldesktop.jar (as done with it's native lib already)
|
| |
|
|
|
|
| |
exception handling in NWReflection and NativeLibLoaderBase.
goal was to prevent catch(Throwable) and to fix or document empty catch blocks where it makes sense.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- See https://bugzilla.mozilla.org/show_bug.cgi?id=486277
- Calling XCloseDisplay occasionally leads to a SIGSEGV,
even thought the reference is valid and OK.
Workaround is not to close any X11Display,
but to hold them stashed and reuse them.
Since we already pipeline all X11Display's
via Nativewindow's X11Util, an added referenceCounter
and a global active/passive list solved this problem.
This workaround is only active in case 'isVendorATI()'.
NEWT/NativeWindow X11:
- Let XIOErrorHandler and invalid display references
fail hard with FatalError, otherwise we won't see
the stack trace - and those bugs are indeed fatal.
NativeWindow X11:
- Install XIOErrorHandler, which stays active.
- X11Util.X11Display:
- Add reference counter
- Add global active/passive list.
Passive if reference count == 0
and marked as 'un-closeable' (-> ATI).
Reusing passive members when create a new display.
-
JOGL:
- Use DeleteLocalRef() calls to free temp NIO buffer
in manual *Copied implementation.
- GLDrawableFactoryImpl: Be serious about the shutdown() semantics
- *GraphicsConfiguration:
- Fix the invalid Onscreen/PBuffer/Pixmap determination (X11/EGL/WGL)
- Just return null if not valid
- X11GLXGraphicsConfigurationFactory - FBConfig
- Determine recommendedIndex properly ..
- Don't bail out if a FBConfig is invalid ..
- Use Chooser in case nothing is recommended ..
- X11OffscreenGLXDrawable fixes bugs:
- wrong (int) cast of parent window in XCreatePixmap call
- setting display to zero too early in destruction, ie
before XCloseDisplay
- X11GLXDrawableFactory is using [singleton] shared dummy resources for
- Screen, Drawable and Context
which are utilized in case they are needed ..
They are removed at shutdown call
- GLXVersion gathering in GLXUtil now ..
- DefaultGLCapabilitiesChooser: Respect PBuffer selection
Tests:
- Add DrawableFactory shutdown()
- Add various Offscreen Capabilties
- Add Offscreen and non-pbuffer case
- JUnit Passed (Linux64bit: NVidia/ATI)
- demos.jrefract.JRefract passed (Linux64bit: NVidia/ATI)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- See https://bugzilla.mozilla.org/show_bug.cgi?id=486277
- Description:
- To use PBuffer, a context must be current
- X11Display cannot be switched while using the PBuffer
[within one thread]. Hence we shall try harder to reuse
_the_ user configured X11Display - whenever possible.
This is actually a good thing, ie cleanup up our
code again.
- Changes to workaround/cleanup:
- GLDrawableFactory* methods 'canCreate*()'
are changed to 'canCreate*(AbstractGraphicsDevice)'
to allow pipelining the X11Display.
This reduces the overhead of using a local TLS X11Display.
- WindowsDummyWGLDrawable cstr gets the GLProfile as a parameter now,
this is done while adding X11DummyGLXDrawable - forseeing the
usecase to query available GLProfiles at startup.
- X11DummyGLXDrawable added, following the WindowsDummyWGLDrawable path
to have a dummy GLContext current to fix the ATI bug.
NativeWindow X11:
- Add XIOErrorHandler to identify the fatal failure
of closing a Display (-> ATI bug).
Build:
- Adding ant.jar and ant-junit.jar to the junit compile/run classpath
-
Misc:
- Fix: CreateDummyWindow(..) returns a HWND, not a HDC
- mapToRealGLFunctionName: Added mapping for X11/GLX.
- X11GLXGraphicsConfigurationFactory: Uncommented dead code 'createDefaultGraphicsConfigurationFBConfig'
Tests: Passed (Linux64bit: NVidia/ATI)
Todo: More tests on ATI, especially multithreading/X11Display usage.
|
| |
|
