GitLab

Apparently, glBindAttribLocation() does not work as planned with AMD GPUs. Therefore, all Shaders yielded black images. Replaced all attribute locations with hard-coded ones in the vertex shader and removed corresponding convenience function from cgt::Shader.

refs #382

* cleaned up cgt::Texture's constructors: only two left - to create an empty texture or a non-empty one
* cgt::Texture will no longer hold a copy of the image data in local memory
* removed a lot of redundant/confusing methods
* no longer needed to call uploadTexture() even though you don't want to upload sth.
* Adapted all known code to the new interface
* Removed cgt::TextureReaderDevil

refs #613

refs #386

refs #386

INVALID_RESULT, INVALID_PROPERTIES, INVALID_SHADER is validated automatically by AbstractProcessor::process(). Hence, there is finally no need anymore to validate these three different levels in each processor.

To now, each property hat an _invalidationLevel field that was evaluated by processors when the property had changed in order to determine what has to be done. However, since properties could also be owned by other classes, this design was semantically misleading.
Therefore, it was removed with this commit and replaced by the invalidation map of each processor. Now, this per-processor mapping of property -> invalidation level is managed by the processor itself. Furthermore, the invalidation level is no longer setup during property creation but during AbstractProcessor::addProperty(), which also makes much more sense.

ATTENTION: Due to these intrusive API changes, the code of all processors and other classes handling properties needs to be changed. As a reminder, the implementation of addProperty() also now takes a reference instead of a pointer, so that old code does no longer compile.

refs #542

* Added assertion in ImageRepresentationGL constructor checking for matching number of channels

* Updated GlGradientVolumeGenerator to generate GL_RGB16F textures to save memory

* Fixed SliceExtractor not rendering Y slice position correctly in XZ-plane mode
* Minor improvements in DataContainerInspector fragment shader and GlGradientVolumeGenerator

C++ implicit conversion rules made ShaderManager::loadSeparate() ambiguous in some cases. So I decided to cut loose ends and completely refactored loading with standard version into ShaderManager::load() and loading with custom version into ShaderManager::loadWithCustomGlslVersion().
ShaderManager::loadSeparate() is no longer available!

Refactoring AbstractProcessor::process() for clearer semantics and better and more uniform handling of invalidation levels:
 * AbstractProcessor::process() now calls updateShader(), updateProperties() and/or updateResult() with respect to the current invalidation level
 * each processor shall no longer override process() but the updateXYZ() methods, at minimum updateResult()
 * AbstractProcessor::process() takes care of (un)locking the processor itself (no need to do this from the outside anymore)

Further implicit changes:
 * Removed redundant HasPropertyCollection::updateProperties()

* Moving to Apache 2.0 license
* Updated AUTHORS.txt

* Split up ProcessorDecoratorShading in ProcessorDecoratorShading inheriting from ProcessorDecoratorGradient
* Introducing GlGradientVolumeGenerator processor in preprocessing module