2. Install BPS from the bundle#
This is the recommended path for scientific users who want a working BPS without modifying processor code. Installs all 14 processors at once from the Aresys-distributed conda bundle.
Download the BPS bundle#
Request access if you do not have credentials yet (free, account-based)
Download the bundle matching the version you want, e.g.
bps-bundle-v4.4.3.tar.gzPlace it in
~/bps-work/SW/:mv ~/Downloads/bps-bundle-v4.4.3.tar.gz ~/bps-work/SW/ mkdir -p ~/bps-work/SW/BPS_V443 tar -xzf ~/bps-work/SW/bps-bundle-v4.4.3.tar.gz -C ~/bps-work/SW/BPS_V443
Run the install notebook#
Open
notebooks/0_BPS_installation.ipynb
in JupyterLab, VS Code, or any notebook front-end.
Set the BPS version at the top of cell 0:
BPS_VERSION = "4.4.3" # match your tarball
If your working directory is not /home/jovyan/, also edit the path
constants near the top of cell 0 to point at your local layout
(~/bps-work/SW/, ~/bps-work/BPS/docs/tutorials/run-bps-locally/).
Run all cells. The notebook:
Creates a conda environment (e.g.
BPS_443) with Python 3.12Builds a local conda channel from the bundle packages
Installs all BPS processors (
bps-l1_processor,bps-l1_framing_processor,bps-stack_processor,bps-l2a_processor, …)Patches
set_environment.bashwith the correct pathsAdds a shell alias to
~/.bashrc(or~/.zshrc):alias bps443='source ~/bps-work/BPS/docs/tutorials/run-bps-locally/CONFIGURATION_FILE/set_environment.bash'
Verifies the installation by running
--helpon every processor
Verify#
After install completes, open a fresh terminal and check:
bps443
which bps_l1_processor
bps_l1_processor --help
If installation fails
A failed install often leaves a partially created conda environment, which causes subsequent retries to fail. Clean up before retrying:
conda env list # 1. confirm the broken env exists
conda env remove --name BPS_443 # 2. remove it
conda env list # 3. confirm it is gone
Then re-run the install steps. If it fails again at the same point, the most likely causes are insufficient RAM (the conda solver crashes when resolving all packages at once) or a corrupted bundle.