See https://scipp.readthedocs.io for project context and how to use and install the scipp Python module.

It is not necessary to build scipp from source if only the Python package is required. See installation on how to install using conda or docker instead.

XCode 10.2 or greater provides a clang++ implementation with sufficient language support for scipp.

You will need to be using LLVM Clang version 7 or greater.

You will need version 7 of GCC as a minimum.

To build and install the library:

git submodule init
git submodule update
mkdir -p build
mkdir -p install
cd build
cmake -DPYTHON_EXECUTABLE=/usr/bin/python3 -DCMAKE_INSTALL_PREFIX=../install ..
make -j4 all-tests install

Note that the all and install targets do NOT include the tests. Those must be built separately via all-tests. You should use Python 3.5 or greater.

To use the scipp Python module:

cd ../python
python3 -m pip install -r requirements.txt
export PYTHONPATH=$PYTHONPATH:../install

In Python:

import scipp as sc

To run the C++ tests, run (in directory build/):

./units/test/scipp-units-test
./core/test/scipp-core-test

Note that simply running ctest also works, but currently it seems to have an issue with gathering templated tests, so calling the test binaries manually is recommended (and much faster).

To run the Python tests, run (in directory python/):

cd python
python3 -m pip install -r requirements.txt
python3 -m pytest

The demo/tutorial notebooks are using a plugin from nbextension, which needs to be installed separately:

pip install jupyter_contrib_nbextensions
jupyter contrib nbextension install --user

pip install jupyter_nbextensions_configurator
jupyter nbextensions_configurator enable --user

Use pip3, depending on your environment.

1. -DDYNAMIC_LIB forces the shared libraries building, that also decreases link time.

For development purposes -DCMAKE_INTERPROCEDURAL_OPTIMIZATION=OFF -DDYNAMIC_LIB=ON is recommended.