View on GitHub

skelebot

Machine Learning Project Development Tool

Home


Dependencies

For both R and Python projects, Skelebot offers the ability to list your dependencies directly in the skelebot.yaml file and have these dependencies installed into your Docker image automatically.

...
dependencies:
- data.table=1.11.2
- stringr
- ...

By default R dependencies are installed using install.packages from CRAN.

By default Python dependencies are installed using pip install.

Versions for packages in R can be specified by appending ={version} to the end of the dependency name.

Versions for packages in Python can be specified by appending ={version} or =={version} to the end of the dependency name.

R and Python also both support dependencies to be installed from the local file system as well as from GitHub using the following structure.

Python also allows for installs using a text file via req:requirements.txt syntax or using a pyproject.toml via proj:pyproject.toml syntax. When using the later, all specified required and optional set(s) of dependencies will be installed.

language: R
dependencies:
- {type}:{source}:{name}
- file:libs/myPackage.tgz:mypack
- github:myGitHub/fakeRepo:fakeRepo
language: Python
dependencies:
- {type}:{source}
- file:libs/myPackage.tgz
- req:requirements.txt
- github:myGitHub/fakeRepo
- proj:pyproject.toml

CodeArtifact Python Packages

Skelebot also supports pulling Python packages that are stored in AWS CodeArtifact. This requires a good deal of information in order to authenticate and pull the correct asset for the package.

language: Python
dependencies:
- ca_file:{domain}:{owner}:{repo}:{pkg}:{version}:{profile}

These values are needed in order to preperly obtain the package and include it in the Docker image.

NOTES:


<< Docker Host | Image Commands >>