the filename extension is "vert" or "frag" but still lets this be
overridden by the Options (for those crazy people who store their
fragment shaders in .vert files :) )."
ADD_LIBRARY(${TARGET_TARGETNAME} MODULE ${TARGET_SRC} ${TARGET_H})
which gives them .so extensions. Since ".so" != ".dylib" osgDB::listAllAvailablePlugins finds no plug-ins. I believe the correct solution is to use CMAKE_SHARED_MODULE_SUFFIX instead. This builds and runs correctly on OS X but I have not tested on other platforms.
Attached is an updated src/osgDB/CMakeLists.txt based on rev 9915. The change is at line 108. To validate: build and then run bin/osgconv --formats. You should get many screenfuls of plug-in features, extensions and options."
In my example case, there are 2 threads - one is a worker thread created by OpenThreads::Thread. The other thread is the main thread i.e. the thread that is intrinsically created when you execute the application. The crucial problem is that for the main thread, OpenThreads::Thread::CurrentThread() will return null.
I'll demonstrate this by breaking ReentrantMutex::lock() into sub-statements:
1.) if (_threadHoldingMutex==OpenThreads::Thread::CurrentThread())
2.) if (_lockCount>0){
3.)
OpenThreads::ScopedLock<OpenThreads::Mutex> lock(_lockCountMutex);
++_lockCount;
return 0;
4.)
int result = Mutex::lock();
if (result==0)
{
OpenThreads::ScopedLock<OpenThreads::Mutex> lock(_lockCountMutex);
5.)
_threadHoldingMutex = OpenThreads::Thread::CurrentThread();
_lockCount = 1;
return result;
An error will occur in the following case:
1) The worker thread calls lock(), it gets to the start of statement 5.
2) The main thread calls lock(). Statement 1 is evaluated as true as _threadHoldingMutex is null, and OpenThreads::Thread::CurrentThread() returns null.
3) The worker thread executes statement 5.
4) The main thread executes statement 2 and evaluates it as true, because the worker thread has set _lockCount to 1. The main thread executes statement 3, and now can access the mutexed-data at the same time as the worker thread!
The simple solution to this is to always protect access to _lockCount and _threadHoldingMutex using _lockCountMutex. I have done this in the file I am submitting."
Added handling of Image::getPixelAspectRatio() in geometry sizing.
Added scaling of the projection matrix to ensure that aspect ratio is honoured with running in fullscreen mode
1. Makes IF/ELSE/ENDIF code blocks easier to read by replacing code like this:
IF(FOO)
MESSAGE("FOO is true")
ELSE(FOO)
MESSAGE(" ??? ELSE(FOO)??? ")
ENDIF(FOO)
with this:
IF(FOO)
MESSAGE("FOO is true")
ELSE()
MESSAGE("FOO is clearly false")
ENDIF()
2. Also adds an ELSEIF() where it makes sense to do so (safe to use as of CMake 2.4.4)"
At the company where I work we model literally thousands of 3DS models for all types of objects. Most don't have transparent textures applied to them, but a few do. The issue is that a texture may have an alpha channel defined, yet its 3DS material definition is not marked as being an alpha source, or transparent. This can happen for any number of modelling reasons (so I'm told) but the net effect is that when a 3DS object is loaded with such a material applied, the stateset creation for the 3DS geometry omits to add the GL_BLEND attribute as neither of the conditions that it currently tests for are satisfied under this scenario.
However, one thing that is apparently an indication of transparency/alpha information that we do make use of when modelling, but that the 3DS reader omits to test, is that we assign to the transparent texture to the opacity_map of the 3DS material structure.
Now, having discussed with our modellers the theory behind the opacity map, it seems that it is safe to say that if an opacity map is specified, then the reader should really apply the GL_BLEND attribute as it is an indication of some form of blending being required between materials.
With this in mind, I have made a minor change to the function createStateSet below, from the ReaderWriter3DS.cpp file, that checks for an opacity map having been specified, and in such a scenario the GL_BLEND attribute is applied to the geometry. This fixed our issue, and thought it might be helpful to others."
When 2 geometries are merged, the primitive sets of the second geometry
are copied to the first geometry.
The primitive sets were copied with a std::insert into the first geometry
primitive set vector. It doesn't work when the geometry is using VBOs (because
the element buffer object of the primitive set is not updated).
The correction replaces
lhs.getPrimitiveSetList().insert( lhs.getPrimitiveSetList().end(),
rhs.getPrimitiveSetList().begin(),
rhs.getPrimitiveSetList().end() );
by
for( primItr=rhs.getPrimitiveSetList().begin();
primItr!=rhs.getPrimitiveSetList().end();
++primItr )
{
lhs.addPrimitiveSet(primItr->get());
}"
> I add META_OSGMANIPULATOR_Object macro which define className, libraryName,
> isSameKindAs methods.
> Clone method is not appropriate for osgManipulator Object."
Node::Node(Node &node, copyop) :
_stateSet(copyop(node.getStateSet()),
It doesn't call the setStateSet method of osg::Node (or osg::Drawable). So the parent
list of the state set is not updated with the new node (drawable)."
some stuff out into DarwinUtils.h/.mm so both implementations can share
some of the code. There's even a bugfix for GraphicsWindowCarbon, which
fixes some issues with multiple windows on different screens."
On destruction of some static variables, the global referenced mutex is used
to lock access to the parent lists of state attributes, nodes and so on.
This even happens past the mutex is already destroyed.
This change to Referenced.cpp revision 9851 uses the same technique like the
DeleteHandlerPointer already in Referenced.cpp to return an zero pointer for
the global referenced lock if it is already destroyed."
"There is error in WoW shader, you can see it by this simple example:
osgviewer cessna.osg --wowvx-42 --clear-color 0,0,0
Clear color may be choosed any with at least one component equals to 0
or 1. In my case I see weird blinking between normal image and image
with depth map at right side on the screen."
Arbitrarily zooms 10% in/out for each click."
Note from Robert Osfield, flipped the orienation of the zoom to make the right mouse key zoom and the scroll wheel work in the same direction.
Single features are implemented as bits asother enums in ReaderWriter class already does, so that combinations are possible and fast comparison operations are possible
By default all features are enabled.
I have added this virtual method to the dot plugin to get an idea how to use these features.
With this patch osgconv --formats shows an additional line 'features' for each plugin"
Because there already exists the option to convert slow path geometry to the fast path by computing an internal fast path alternative, I added a new optimizer option that automatically does this. To check the results I also made some changes to the statistics gathering and rendering.
Somewhat unrelated, but also part of the optimizer I disabled removal of CameraView nodes during RemoveRedundantNodes optimization.
As discussed on the ML, CameraViews were removed from the scenegraph. This solves that issue.
Summary:
-Geometry::areFastPathsUsed now also looks at internalOptimizedGeometry
-Added Optimize option to make all slow path geometry compute their internal fast path alternative
-Added fast geometry counter to the statistics
-Disabled removel of CameraViews in optimizer
"
"Attached you'll find a proposal for using different
protocols. The idea behind the new code is:
1.) plugins/apps register protocols which they can handle. This is done
via osgDB::Registry::registerProtocol(aProtocolName). Plugins register
supported protocols as usual via ReaderWriter::supportsProtocol(..), the
Registry is updated accordingly.
2.) osgDB::containsServerAddress checks first for an appearance of "://"
in the filename and then checks the protocol against the set of
registered protocols via Registry::isProtocolRegistered(aProtocollName)
3.) the other getServer*-functions changed as well, there's even a
getServerProtocol-function
With these changes filenames/Urls get routed to loaded plugins even with
different protocols than 'http'."
osgText::Text and osgText::Text3D use the same font file.
The first really load the file and obtain an osgText::Font object,
the second use the cache created during the first load of the
font file, and so obtain an osgText::Font object instead of
osgText::Font3D object. To obtain an osgText::Font3D object,
osgText::Text3D call osgDB::readObjectFile(...) with an option
to specify the plugin we want an osgText::Font3D instead of
osgText::Font.
Generalised Problem:
In osgDB::Registry, loaded file cache is referenced by the name
of this file, so if I load a file with some options, and the cache
already contain object for this filename, I obtain an object
potentially not loaded with my options.
Behaviours:
Cache management is delegate to osgDB::Registry, but cache
coherence (load a file with option then reuse it, deactivate the
cache when load a specific file or don't cached the loaded file)
is user's responsibility.
Text3D solution:
Postfix the font file name by .text3d or something similar and then have the freetype plugin return
osgText::Font3D when it detects this.
This operation is done by osgText::readFont3DFile() which unsure the filename have .text3d as extension.
This is totaly transparent for user, and backward compatible.
BTW, I fix the bug about the Normal of 3D text. Currently, the front and wall face have
the same normal (0,0,1) in the Text3D object coordinate. Now the wall face have its own
normal array computed by the plugin.
BTW 2, I implement
- void Text3D::accept(osg::Drawable::ConstAttributeFunctor& af) const
- void Text3D::accept(osg::PrimitiveFunctor& pf) const
so now statistics are well reported.
"
