Agilepy-recipe

This repository containes the recipes to build the Docker container of Agilepy.

For Docker :

Building the Docker images

Base image

This Dockerfile starts from an oracle8 image and installs all the low and high level dependencies required by Agilepy. The AGILE tools are installed from source. It can be used as develop environment.

Build the base image and push it to DockerHub. Tags here

cd recipes/docker/base
docker build --tag agilescience/agilepy-recipe:<tagname> .
docker login
docker push agilescience/agilepy-recipe:<tagname>

Environment updates

Agilepy's dependencies are locked to provide a stable environment. When a new dependnecy is added to the requirements.txt, the requirements.lock file must be regenerated with:

pip-compile --resolver=backtracking --no-annotate --output-file requirements.lock requirements.txt

pip-compile (>=6.12.3) can be installed with

pip install pip-tools

If you want to update a package, regenerate the .lock file using the --upgrade-package PACKAGE option.

Agilepy image

Add to the base image a specific release of Agilepy and push it to the DockerHub. Agilepy releases here

cd recipes/docker/agilepy
docker build --build-arg BASE_VERSION=<base-version> --build-arg AGILEPY_RELEASE=<agilepy-release> --tag agilescience/agilepy:release-<agilepy-release> .
docker login
docker push agilescience/agilepy:<agilepy-release>

Building the conda packages (deprecated)

For Anaconda the following packages are available on the Anaconda Cloud:

agiletools

This package contains the AGILE science tools. They are downloaded and installed using the AGILE-GRID-ScienceTools-Setup repository.

agilepy-dataset

This package contains two datasets, extracted from the AGILE data archive. They are used within the Agilepy software for testing purposes and for developing several Agilepy's tutorials.

agilepy

This package contains the Agilepy software. The agiletools version in meta.yaml must be updated to the latest one available on Anaconda Coud.

Uploading the package

anaconda login

Upload the package with:

anaconda upload <path-to-package>

The is written on the console at the end of the conda-build command.

Update conda and install and update conda-build:

conda update conda
conda install conda-build
conda update conda-build

Add the required channels.

conda config --add channels conda-forge
conda config --add channels plotly

Build the anaconda packages:

source recipes/anaconda/science-tools/start_build.sh <agile-st-tag>
source recipes/anaconda/agilepy-dataset/start_build.sh <agile-st-tag>
source recipes/anaconda/agilepy/start_build.sh local <agilepytag>

Troubleshooting

The command anaconda login is not found

Be sure to use the anaconda Command line client.

anaconda --version
anaconda Command line client (version 1.7.2)

and not

anaconda --version
anaconda 21.48.22.156-1

AGILESCIENCE/Agilepy-recipe