jakekonrad / pyglzip

Graph compression techniques for graph learning systems

glzip

Graph compression for graph learning systems

Building from source/Developers

Building the environment

Conda commands to recreate the virtual enviroment I am using

conda env create -f ${CUDA}/environment.yml
conda activate glzip_${CUDA}
pip install -r ${CUDA}/requirements.txt

where ${CUDA} should be replaced by cpu, cu102 or cu113 depending on your device.

Updating the environment

If you need to make changes to the environemnt. First clear the environment:

conda deactivate
conda env remove -n glzip_${CUDA}

then rebuild it.

Building glzip

With the environment active run maturin develop. That will install the glzip module into the environment. maturin develop --release will build the module with optimizations turned on which makes everything run much, much faster so if you are not developing the library and do not need debug symbols, I recommend building with --release.

Updating glzip or making changes to the source.

Just run maturin develop or maturin develop --release again after making the changes.