Docker

About

Having functional containers is essential for efficient CI and for ensuring a reliable environment conducive to reproducibility of research. There are a few versions of the LuSEE software Docker images, targeted towards the development of the luseepy Python package, its testing environment, CI Jupyter notebooks etc. These images are hosted on Docker Hub.

Singularity

Docker is not available in the interactive nodes of SDCC. However, it is possible to run Docker images with Singularity.

This is an example of a Singularity session using the “base” Docker image which contains all the packages needed by the luseepy package. The luseepy package in this case is cloned from GitHub and its folder is used in bind mode in Singularity i.e. it is available for editing and other modifications on the host machine.

# Create a working directory
$ cd /home/myhome/ # replace with desired locaiton

# Clone luseepy
$ git clone git@github.com:lusee-night/luseepy.git

# Start Singularity. Note:
# a) We pull the "base" image from Docker Hub, from the repository maintained by the user "buddhasystem"
# b) We "bind" the local lusee folder to "/user" accessible within the container
$ singularity exec --bind /home/myhome/luseepy:/user docker://buddhasystem/lusee-night-luseepy-base:0.1 bash

# We are now running an interactive "bash" shell. Let's set up the PYTHON environment:
Singularity> cd /user/lusee # remember that the local "luseepy" folder is bound to "/user"
Singularity> source ./setup.sh

# Run a test of the "LunarCalendar"

Singularity> cd ../tests
Singularity>  ./LunarCalendarTest.py 
2025 04 14 19 59 28 753901
2025 05 12 06 28 21 606078

At this point the user can do any sort of modifications to the files in the lusee folder, for example by opening an emacs session in a separate window (since emacs is not available in the base lusee image).