GitLab

Qt threading issues

Lua console

Finished a quick and dirty implementation of a Lua console in CampvisApplication.

* CMake build files now support unix builds
* SWIG interface files and Lua wrappers are now fully C++ compliant to comfort GCC
* Hacked shared/static builds of the campvis-modules module even more, but now it works on both MSVC and GCC

* AbstractData now stores a weak_ptr to its shared_ptr owning group if existant. This avoids unintentional creation of multiple owning groups where each eventually tries to delete the AbstractData on its own (which will certainly not work)
* Updated DataContainer implementation to use a hash map instead of an unordered map.

Enforced update of min/max/singleStep values in AbstractAdjusterWidget to be performed in Qt main thread through additional indirection via Qt queued signals.

Refactor reference counting

* Introduced two new tgt::LogManager log levels LuaInfo and LuaError.
* ScriptingWidget registers itself as logger and prints the Lua output into the console window.

* Added OrCombinedPointPredicate implementation
* Updated default settings in PredicateDemoCarotid2 pipeline

(Apparently, I really need some rest...)

Itk reader

ITK Image reader that supports many more image formats.
Example usage edited into the itkfilterdemo pipeline.

 * supports all file formats that can be read via ITK
 * does not support image series

* SliceExtractor FitToWindow property defaults to true to mimic old behavior per default
* PointPredicateRenderArea renders y axis label

* Now supports optional scaling and offset of image
* Added properties for geometry render style and size
* Fixed some scribbling-related bugs

Slice extractor scribbe fix

as discussed

Datastructures

basic datastructures for Matrix and position data

Slicerenderer2d

slice renderer that applies transfer function, clips image and/or inverts axes.
tested only with grayscale input images, but should support color and images as well.

* undo changes to .gitignore (regarding SWIG temporary files)
* moved implementation of PositionData and TransformData from header to source file

Mhdreader fix

Sorry for the huge commit - only couple of files changed (see commit) :-(

* Implemented one single gloabel Lua VM for the entire CampVisApplication instead of one VM for each pipeline.
* The global Lua VM has a "pipelines" table/array, holding the pointers to each loaded pipeline
* ScriptingWidget support cycling through last executed commands via the arrow keys
* LuaVmState supports redirecting Lua's print() function to a custom one that uses tgt::Logmanager for printing (just as proof-of-concept). This shall later be extended to pass all Lua output to the scripting console.

With scripting enabled, the main window now has an additional scripting widget that allows to interact with a Lua VM. The current proof-of-concept implementation creates a Lua VM for every instantiated CAMPVis pipeline while the console just interacts with the first pipeline's one.
However, the Lua VM concept has to be reiterated later anyway. It may make more sense to have just a single central Lua VM for the entire application, which is shared by each pipeline.

Fixed timing issue in CampVisPainter that led CampVisPainter::paint() being called before CampVisPainter::init(). CampVisPainter::paint() now ensures the class (and in particular the shader) to be correctly initialized.

fixes #564

Initial implementation of a Lua scripting layer

Over the last several months a scripting layer that allows pipelines to be defined in Lua has been developed. It uses SWIG to generate Lua modules with bindings for CAMPVis classes. It is an opt-in feature, and tries to be as non-intrusive to standard CAMPVis code as possible.

The implementation of the scripting layer has reached a state where it's possible to write fully-functional pipelines in Lua. In fact, 2 existing pipelines have been reimplemented in Lua for testing purposes and added to the project. In my opinion, this marks a good point to merge the initial implementation into the development branch — that would make it easier to test and improve it.

Naturally, there are still many rough edges that should eventually be dealt with, but they can be addressed separately as new features:
- bindings coverage is rather low
- Lua pipelines currently need to be statically registered
- pipeline definition syntax could be streamlined (e.g. by getting rid of the `instance` global variable)

The CMake-enabled version of Lua that CAMPVis' scripting layer uses
hasn't been updated in a while. Synchronise it with upstream now that
we're getting ready to merge the `swig` branch in.

References #1

`scripting/testapp.cpp` contains a simple console application that was
used to test CAMPVis' scripting layer before it was ready for
integration with the main GUI application. Now that Lua pipelines can be
registered and executed the same way regular ones are, the test
application is no longer needed. Consequently, this commit removes it
from the project.

References #1

Glue classes that live in the `scripting/glue` directory were mostly
undocumented, and the documentation of LuaPipeline was incomplete.

Improve the documentation of LuaPipeline and LuaVmState. Document
LuaTable, RegularLuaTable and GlobalLuaTable from the `scripting/glue`
directory.

References #1

Starting with commit d6fec679, pipelines need to provide their own light
source to function properly. Lua pipelines didn't, which caused them to
fail to render.

Add a SWIG wrapper for LightSourceProvider to make it accessible from
Lua. Instantiate and attach a light source in existing Lua pipelines.

References #1