Skip to content

Latest commit

 

History

History
155 lines (130 loc) · 5.72 KB

BUILDING.md

File metadata and controls

155 lines (130 loc) · 5.72 KB
Raw

Building Tomviz

The Tomviz project reuses a number of components from Python, VTK, ITK, and ParaView, so these dependencies will need to be compiled first in order to build Tomviz. There is a superbuild that automates most of this when building binaries. We recommend using these binaries where available if you wish to test, and nightly binaries are made available in addition to releases.

To develop Tomviz we recommend using our developer superbuild. That repository will clone this one, and several others such as the correct ParaView version and optionally the ITK release needed. It will build them all with the correct flags, and ensure they link to the correct versions of projects. Going forward this is our preferred development build environment, enabling a quick and easy set up. The use of git submodules ensures you will not lose any changes and will have full access to the source code. You should commit changes from the Tomviz repository contained there.

Advanced Build Instructions

These are retained for historical reasons and for the brave. Most of these steps are contained and automated in the tomvizsuper repository, and that also tries to ensure that the same Qt, Python, etc are used in the contained projects. This is how you pass the flags circa late 2019, and in early 2020 we decided to remove some automated build consistency checks that mean developers must work harder to ensure build consistency with respect to Qt, Python and other third party libraries.

If you wish to develop Tomviz, the following sequence of commands will build a suitable ParaView, ITK, and a Tomviz that uses these builds. You should use your package manager/installers/upstream build instructions to install the dependencies. CMake will search for these, and the cmake-gui can be used to point to them if they are not found automatically.

Dependencies

  • Qt 5.9+ (5.9 recommended)
  • CMake 3.3+
  • Python 3.7
  • NumPy 1.12
  • Git 2.1
  • C++ compiler with C++11 support (MSVC 2015 on Windows)
  • Intel TBB

Initial Build

You will need the development headers, so please ensure they are installed (several distros use -dev packages not installed with the main package). If you are in a directory where you would like to place the source and builds with all prerequisites installed:

git clone --recursive git://github.com/openchemistry/paraview.git
mkdir paraview-build
cd paraview-build
cmake -DCMAKE_BUILD_TYPE:STRING=RelWithDebInfo \
  -DBUILD_TESTING:BOOL=OFF \
  -DPARAVIEW_USE_PYTHON:BOOL=ON \
  -DPARAVIEW_ENABLE_WEB:BOOL=OFF \
  -DPARAVIEW_ENABLE_EMBEDDED_DOCUMENTATION:BOOL=OFF\
  -DPARAVIEW_USE_QTHELP:BOOL=OFF \
  -DPARAVIEW_PLUGINS_DEFAULT:BOOL=OFF \
  -DVTK_SMP_IMPLEMENTATION_TYPE:STRING=TBB \
  -DVTK_PYTHON_VERSION:STRING=3 \
  -DVTK_PYTHON_FULL_THREADSAFE:BOOL=ON \
  -DVTK_NO_PYTHON_THREADS:BOOL=OFF \
  ../paraview

Check which version of Python CMake found. Ensure the PYTHON_EXECUTABLE, PYTHON_INCLUDE_DIR, and PYTHON_LIBRARY variables are pointing to a consistent version of Python in the same prefix. Compiling against Anaconda packages is not recommended. If they are consistent proceed to build ParaView:

cmake --build .

This will clone all source code needed, configure and build a minimal ParaView, the Ninja generator is recommended for faster builds, but optional. On Windows you will need to specify the correct generator for the installed compiler.

cd ..
git clone git://itk.org/ITK.git
cd ITK
git checkout v5.0.1
cd ..
mkdir itk-build
cd itk-build
cmake -DCMAKE_BUILD_TYPE:STRING=RelWithDebInfo \
  -DITK_LEGACY_REMOVE:BOOL=ON \
  -DITK_LEGACY_SILENT:BOOL=ON \
  -DModule_ITKBridgeNumPy:BOOL=ON \
  -DBUILD_TESTING:BOOL=OFF \
  -DITK_WRAP_unsigned_short:BOOL=ON \
  -DITK_WRAP_rgb_unsigned_char:BOOL=OFF \
  -DITK_WRAP_rgba_unsigned_char:BOOL=OFF \
  -DITK_BUILD_DEFAULT_MODULES:BOOL=OFF \
  -DITKGroup_Core:BOOL=ON \
  -DITKGroup_Filtering:BOOL=ON \
  -DITKGroup_Segmentation:BOOL=ON \
  -DITKGroup_Nonunit:BOOL=ON \
  -DPython_ADDITIONAL_VERSIONS:STRING=3 \
  -DITK_WRAP_PYTHON:BOOL=ON \
  -DBUILD_EXAMPLES:BOOL=OFF \
  -DBUILD_SHARED_LIBS:BOOL=ON \
  ../ITK
cmake --build .

This will build ITK with its Python wrapping turned on, you must inspect the CMake configuration to ensure it uses the same Python as ParaView. ITK, while not required to build Tomviz, is required for any of the ITK-based data operators to run.

Now, to build Tomviz:

cd ..
export build_root=`pwd`
git clone --recursive git://github.com/openchemistry/tomviz.git
mkdir tomviz-build
cd tomviz-build
cmake -DCMAKE_BUILD_TYPE:STRING=RelWithDebInfo \
  -DParaView_DIR:PATH=$build_root/paraview-build \
  -DITK_DIR:PATH=$build_root/itk-build \
  ../tomviz
cmake --build .

Running Tomviz

You should be able to simply execute the binary from the build tree, on Windows this is a little more difficult as you must ensure all DLLs are in the same directory or in your 'PATH' environment variable.

./bin/tomviz

Updating Source and Rebuilding

Once this completes you will have a binary in the 'bin' directory of your build tree. It is possible to update the ParaView and/or Tomviz source directories and rebuild the binaries incrementally. The following would update both source trees and rebuild the latest version of each:

cd ../paraview
git pull && git submodule update --recursive
cd ../paraview-build
cmake --build .
cd ../tomviz
git pull && git submodule update
cd ../tomviz-build
cmake --build .