v3.9.0

• VM: set the matplotlib backend to tkagg
• Updated Versions:
  • CIL: numpy-2 branch (v25+)
  • STIR: 6.3.0
  • SIRF: 3.9.0
  • ITK: 5.4.4
  • parallelproj: 1.10.2
  • boost: 1.85.0 (maximum supported by Gadgetron)
  • CCPi-Regularisation-Toolkit: v25.0.0
  • SWIG: 4.4
• Docker:
  • separate build & test steps
  • fix missing site-packages
  • STIR_DISABLE_HDF5=ON pending future fix (see #980)
• CMake:
  • Use new FindPython module. You might need to help it find the correct Python path, e.g. by setting Python_ROOT_DIR.
  • Use IPP_ROOT instead of IPP_LIBRARY (for BUILD_CIL)
  • Suppress compilation warnings for Gadgetron for future compatibility (as Gadgetron converts them to errors)
  • Support the Gadgetron_EXTRA_CMAKE_ARGS option
  • Add SIRF_DISABLE_Gadgetron_toolboxes convenience flag
  • fix HDF5 problems related to finding C/CXX components & version restrictions
  • update & cleanup CIL tests, add unittest-parametrize
  • fix ISMRMRD build (c++17, cURL docker patch)

v3.8.0

• Removed ACE as no longer needed by Gadgetron

• Updated Versions:

  • SIRF: 3.8.0
  • STIR: 6.2.0
  • parallelproj: 1.9.1 (contains a fix for truncation of the TOF kernel)
  • CIL: v24.1.0
  • Armadillo: 12.8.4 (as per latest Gadgetron, see #923)
  • pet-rd-tools: 2.0.2

• docker image updates:

  • fix CUDA toolkit on GPU image
  • use CMake 3.29.6

• VM updates:

  • use CMake 3.29.6

v3.7.0

• Adds CCPi-Regularisation-Toolkit as prerequisite for CIL and reinstate unit tests
• Docker:
  • Handle BUILD_CIL correctly and not install CIL from conda if BUILD_CIL=ON
• CMake:
  • only pass through OpenMP*LIBRARY variables if set, and add OpenMP_iomp_LIBRARY to the list (for Intel compiler) #879
• Updated Versions:
  • SIRF: v3.7.0
  • CIL: ccf17f393ba911d13b74f2327779dde030098fe6 # 28 May 2024
  • CCPi-Regularisation-Toolkit: v24.0.1
  • STIR: v6.1.0

v3.6.0

• build the STIR native Python interface by default (STIR_BUILD_SWIG_PYTHON=ON). You can still switch this off.
• docker image updates:
  • uses docker-stacks
  • CPU: ubuntu:22.04
  • GPU: nvidia/cuda:11.7.1-cudnn8-runtime-ubuntu22.04
  • Important: Docker tags have now changed, see docker/README.md for a table. Changes w.r.t. previous versions are listed below (note that -gpu versions of all tags exist as well, e.g. latest-gpu etc).
    • The latest tag now includes the jupyter server etc, so the service tag should no longer be used (and corresponding images on DockerHub will not be updated).
    • The new edge tag corresponds to master branch of the SIRF-SuperBuild. Note that this will include the "normal" versions of all packages.
    • The devel tag (for DEVEL_BUILD=ON with master versions of SIRF, STIR etc) is no longer pushed to DockerHub, and therefore will not be updated.
  • Docker images are now pushed to DockerHub and GHCR
  • added requirements.yml
  • revert to conda but with libmamba solver
• updated versions:
  • SIRF: v3.6.0
  • STIR: v6.0.0
  • parallelproj: v1.7.3
  • Gadgetron: 6202fb7352a14fb82817b57a97d928c988eb0f4b (version of Oct 13, 2023)
  • ISMRMRD: v1.13.7
  • NiftyReg: a328efb3a2f8ea4b47cf0f7b581d983a570a1ffd (version of 8 Mar, 2024, fixing find omp.h problems on MacOS)
  • siemens_to_ismrmrd: v1.2.11
  • CIL: db5a2a6cd3bddfbbf53e65f0549ac206096e5b44 (version of Feb 13, 2024)
  • CCPi-Regularisation-Toolkit: 71f8d304d804b54d378f0ed05539f01aaaf13758 (version of Dec 3. 2023)
  • ROOT: v6-28-12
  • nlohmann/json: 3.11.3
• CMake changes
  • pass through OpenMP_CXX_INCLUDE_DIR and OpenMP_C_INCLUDE_DIR to
    dependent projects. For instance, on MacOS the compiler might not find
    omp.h unless these variables are set.
• VM changes
  • UPDATE.sh (and hence update_VM.sh) has extra developer options to set the
    DEVEL_BUILD=ON (-D) and skip the build (-S).
• Drop Travis

Full Changelog: v3.5.0...v3.6.0

v3.5.0

• Temporarily disable CCPi-Regularisation due to #832
• CMake/building:
  • default DISABLE_MATLAB to ON as SIRF Matlab support is out-of-date and enabling it could
    generate conflicts with Python shared libraries.
• docker image updates
  • introduce REMOVE_BUILD_FILES variable. If set to 1 (which is the default),
    most build files will be removed in the docker image.
  • introduce RUN_CTESTS variable. If set to 1 (which is the default),
    the CTests will be run while building the image.
  • remove obsolete copying of gadgetron.xml
  • Installing requirements for SIRF-Exercises uses its environment.yml or requirements.txt depending on settings.
• Build Gadgetron master. Requires new Ubuntu packages: libdcmtk-dev, libpugixml-dev, libgflags-dev,
  libssl-dev, libcurl4-openssl-dev, pkg-config, golang, libboost-coroutine-dev, libboost-context-dev, libboost-random-dev.
• Added SuperBuild project dependencies for Gadgetron: range-v3, RocksDB, Date, mrd-storage-server
• updated versions:
  • Gadgetron: 42f11bf14b77b16f1ca5bcfbfa435d5ee8cb22a6 (master)
  • RocksDB: 6.26.0
  • range-v3: 0.12.0
  • siemens_to_ismrmrd: 8bb8b08f53ce73c2de9ba5f47f1532f96292d92b
  • ISMRMRD: v1.13.2
  • SIRF: v3.5.0
  • CIL: v23.1.0
  • CCPi-Regularisation: v22.0.0
  • TomoPhantom: v2.0.0

v3.4.0

• Removed CIL-ASTRA as it has been merged into CIL code base.
• docker images updates
  • Ubuntu: 22.04
  • added dependencies to python-opencv , pytest, pytest-cov and coverage
  • added requirements.yml and requirements-service.yml to handle dependencies for conda python
  • updates to use mamba
  • use jupyterlab as opposed to notebook in the "service" images
  • CMake: v3.25.1
• VM updates
  • Ubuntu: 22.04
  • update configuration and documentation to use jupyterlab as opposed to the notebook server
  • CMake: 3.21.3
  • fixes documentation
  • use the VMSVGA graphics controller
  • use environment variables in Vagrantfile for easier building
• updated versions:
  • SIRF: v3.4.0
  • CIL: a6062410028c9872c5b355be40b96ed1497fed2a > 22.1.0
  • GTest: 1.12.1
  • glog: 0.6.0
  • parallelproj: v1.2.13
  • STIR: rel_5.1.0
  • Boost: 1.78.0

v3.3.1

• VM:
  • "update_VM.sh -s" (i.e. "UPDATE.sh -s") no longer runs configure_gnome.sh. If you have a very old VM, run it manually instead.
  • Updates to run using docker scripts
  • installs custom pip and all python prerequisites with pip
  • Bugfix in finding cython and python in UPDATE.sh
  • general refresh of scripts etc
  • move zero_fill.sh from first_run.sh and move it to a new clean_before_VM_export.sh script, which also removes build files to make the exported VM smaller.
• docker and VM:
  • install uuid-dev such that we're prepared for installing ROOT
  • no longer force numpy<=1.20
• CMake:
  • FindCython allows hints

v3.3.0

v3.3.0

• add CITATION.cff (and remove .zenodo.json)
• updated versions:
  • SIRF: 3.3.0
  • STIR: 5.0.2
  • parallelproj: v0.8
  • CCPi Regularisation: v21.0.0
  • CIL: v21.4.1 (CIL devel build to commit hash ef66083de231492f9571f5512b33068f6950e877 )
• Windows
  • Boost build add regex and random
  • install env_sirf.PS1 and .bat files (as the .sh and .csh files)
• docker:
  • fix problems with CUDA repo keys
  • minor fixes to scripts for use elsewhere (including preparation for more recent Ubuntu)
  • added numba as dependency
• VM:
  • set BUILD_CIL=ON
• generic *.cmake fixes:
  • update to the ASTRA build script
  • fix/add minimum versions for various packages
  • improve finding GTest
• known problems:
  • VM and jupyterhub scripts need merging various fixes
  • VM build files are currently not up to date with 3.3.0 version

v3.2.0

• Moved the VM to this SuperBuild repo
• CMake: removed USE_SYSTEM_siemens_to_ismrmrd
• CMake: added required versions for STIR, NIFTYREG, ISMRMRD.
• Drop Python 2 support
• CMake minimum required version is set to 3.16.2, as it supports Boost 1.72
• VirtualBox VM is created from the SIRF-SuperBuild repository
• Docker images install CIL via conda, so setting BUILD_CIL=OFF in the docker build
• Added jupyterhub Dockerfile and documentation to build the jupyterhub image used in training on top of SIRF docker images and jupyter datascience-notebook image with GPU access.
• Adds Boost random for ISMRMRD 1.5.0 #636
• moved the VM repository to the VirtualBox subdirectory
• fix usage of proj_EXTRA_CMAKE_ARGS facility (it was broken for all projects except ITK) #616
• Boost: fix cases where the wrong version of boost could be found #627
• GTest: if USE_SYSTEM_GTest=OFF, attempt to force finding our version, otherwise, pass GTest_DIR or GTEST_ROOT on to dependent projects.
• add the CERN ROOT library (if USE_ROOT=ON, but defaults to OFF), which can be used by STIR to read GATE ROOT files.
• updated versions:
  • CMake: 3.16.2
  • ISMRMRD: 1.7.0 if siemens_to_ismrmrd is built, 1.4.2.1 otherwise
  • siemens_to_ismrmrd: 6d0ab3d3d0c8ade5c0526db1c6af9825008425ad > 1.2.2 with bug-fix for boost/foreach.hpp
  • ITK: 5.2.1 However, we now build a smaller set of modules, most (but not all) of IO, and Filtering. See SuperBuild/External_ITK.cmake)
  • NiftyReg: 99d584e2b8ea0bffe7e65e40c8dc818751782d92 ) (fixes gcc-9 OpenMP problems)
  • CIL: v21.4.0
  • GTest: 1.11.0
  • Boost: 1.72.0
  • JSON: 3.10.4
  • ACE: 6.5.9
  • parallelproj: v0.8

v3.1.0

v3.1.0

• docker:
  • configure nbstripout in SIRF-Exercises for sirf-service
  • major change w.r.t. users and permissions. We now build as user jovyan (by default)
    and still switch to sirfuser for running the container. This avoids having to
    reset permissions of many files, and therefore speeds-up container start-up.
  • introduced SIRF_SB_URL, SIRF_SB_TAG and NUM_PARALLEL_BUILDS. They default to the
    values used before (i.e. resp. the main SIRF-SuperBuild repo, master and 2).
  • add dependencies that are available only from conda-forge (for CIL)
  • improved documentation
• allow specifying HDF5_URL and HDF5_TAG like for other projects
• updated versions:
  • SIRF: 3.1.0
  • CIL: 21.2.0
  • CIL-ASTRA: 21.2.0
• disable built of NiftyPET by default as our current setup requires Python2 for which we dropped support
• Continuous Integration testing:
  • Removed all Travis runs except those that run docker