Build from source

Prerequisites 

Obtain the source code 

To get the source code, the repository should be downloaded with git:

git clone https://gitcode.com/cann/pyasc.git
cd pyasc
git fetch origin +refs/merge-requests/85/head:v2
git checkout v2

Setup the build environment 

Python 3.9 or newer with pip is required for PyAsc.

It is recommended to activate a virtual environment (venv) instead of using a system-wide installation:

python3 -m pip install virtualenv
python3 -m virtualenv $HOME/.pyasc_venv --prompt pyasc
source $HOME/.pyasc_venv/bin/activate

Install build dependencies 

PyAsc requires either gcc or clang compiler and the compatible linker. Optionally, ccache can be installed to speed up the development flow.

Hint

On Ubuntu, these packages can be installed from apt:

sudo apt install build-essential ccache clang lld

Build and install 

One of the approaches below can be used to install or develop the Python package.

The following environment variables can be exported to configure the installation process:

LLVM_INSTALL_PREFIX=<path> (required): provide the directory with pre-built LLVM binaries (preliminarily download and unpack the archive for x64 or arm64 platform)
PYASC_SETUP_BUILD_DIR=<path> (optional): provide the directory for the temporary build files (default: build)
PYASC_SETUP_CCACHE=1 (optional): enable ccache to speed up the repetitive build flow
PYASC_SETUP_CLANG_LLD=1 (optional): use clang and lld instead of the default toolchain
PYASC_SETUP_CONFIG=<config> (optional): build configuration for CMake (default: Release)
PYASC_HOME=<path> (optional): provide the directory that should be used to store pre-built dependencies (default: user home directory)

Also, -v option can be added to pip arguments to increase the output verbosity. This will allow to see e.g. CMake and C++ compiler commands.

Install the package (recommended)

This is the simplest way to build and install PyAsc using a single command. The project repository can be safely deleted after the package is installed into the environment.

The following command should be run to collect build dependencies and install Python package into the environment:

python3 -m pip install .

Note

After the installation, it is recommended to setup runtime environment in order to run PyAsc operators.

Advanced installation options 

Option 1: Install the package (preserve build directory)

This option provides the most setup features for the project developers. It allows to enable development tools.

These environment variables can be exported additionally to configure the installation process:

PYASC_SETUP_DEVTOOLS=1 (optional): build and install development tools
PYASC_SETUP_DOCS=1 (optional): render markdown documentation for IR entities in build/cmake*/docs directory

python3 -m pip install -r requirements-build.txt
python3 -m pip install --no-build-isolation .

Option 2: Install the editable package 

This option is recommended for the project developers since it allows to modify internal Python files without the need to perform the PyAsc re-installation. The --no-build-isolation argument can be added too. However, the development tools will not be available on path automatically when using this approach.

python3 -m pip install -e .

Option 3: Install the distributable package (wheel)

This option can be used if there is a need to obtain the installation archive (wheel) to install the package into another environment or transfer it to other device.

python3 -m pip wheel --wheel-dir=wheels --no-deps .
python3 -m pip install wheels/pyasc*.whl