Skip to content

Conda build

Complete overhaul of CI/CD script and Dockerfiles to use a build image with conda dependencies.

The BayesWave runtime Dockerfile starts from the conda build environment, clones the repo at the commit the CI is triggered by and installs from source.

The CI/CD script also now pushes out a tagged build container, as well as the runtime container, so it will be easier (i.e. possible) to rebuild BayesWave from any tagged point in the past. The old version only ever had a nightly build image.

Most of this should be invisible to most users but note the presence of the environment.yaml: this can be used to create a bayeswave development environment in conda like:

$ conda env create -f environment.yaml
$ conda activate bayeswave
$ git clone https://git.ligo.org/lscsoft/bayeswave.git 
$ cd bayeswave
$ git checkout <some-branch>
$ ./install <some-path>

and should be somewhat faster than replicating the igwn-py3X environment.

Merge request reports