Coder Social home page Coder Social logo

symbiflow-ingen's Introduction

Symbiflow Package Installer Generator (INGEN)

This repo contains a set of scripts which automates the process of creating a symbiflow package installer, also called a symbiflow dailybuild.

This is meant to be run periodically (nightly) on any machine - which will check for updates, create and sanity test a new package installer and then will publish it (if all ok) to github as a release.

The release versions start at v2.2.0 to avoid using any of the older versions which are already in use.
Each release version corresponds to a specific date on which the package installer was published on github.

Developer Notes

  1. use environment.yml and requirements.txt for installing conda environments rather than installing each package separately using conda install <package> because the env solver can easily solve the dependency graph in one go.
    Otherwise, we see issues such as not being able to resolve dependency later, or a step where all the packages are switched between channels for compatibility (conda-forge and defaults for example)

  2. use a conda environment and install the packages directly, rather than using conda search.
    This will use the env solver and we get the right packages as we would get when we actually execute the installer later, conda search depends on the search factors used, and we cannot really specify all the factors easily.
    conda search was not meant to be a user spec driven tool, unfortunately. Once the packages are installed, query their versions and build the installer.

  3. use an existing conda env and update it simply by using a newer environment.yml and requirements.txt rather than deleting the conda env and recreating it with the newer files.

Structure

├── README.md this file

├── ingen_kickoff.sh main wrapper file
├── ingen_generate_package_installer.sh checks for updates and generates a package installer
├── ingen_test_package_installer.sh runs the package installer through sanity (k6n10, k4n8, eoss3, pp3)
├── ingen_publish_package_installer.sh publishes the package installer as a Github Release

├── ingen_conda_helper.sh helper bash script with conda commands wrapped
├── ingen_git_get_commit_list.sh helper bash script to get git repo info

├── ingen_requirements.txt conda (pip) requirements.txt spec for the INGEN env
├── ingen_environment.yml conda environment.yml spec for the INGEN env

├── ingen_generate_package_updates.py helper py script to generate updated dependencies
├── ingen_check_for_updates.py helper py script to analyze generated updated dependencies
├── ingen_generate_conda_env_configuration.py helper py script to generate a conda env for the package installer
├── ingen_generate_installer_script.py helper py script to generate installer script from a template

├── ingen_package_spec.yml base spec to describe every dependency of the package installer
├── ingen_package_current.yml current versions of every dependency in the base spec at the latest release
├── ingen_installer_script_template.sh template for final package installer script

├── installers
├── dailybuild all package installers are available here
├── makeself-2.3.1.run util to create the final self-extracting package installer
└── tests sanity tests for package installer

Workflow

A brief workflow is described, TODO add diagrams/details.

  • [A] generate a new package installer

    • creates a new local conda installation
    • creates a new ingen conda env needed for running rest of the steps (git/jq...)
    • generates requirements.txt and environment.yml for ingen_builder conda env from the ingen_package_spec.yml base specification for the package installer
    • creates a new ingen_builder conda env from the above - this accurately reflects the structure of the 'potential' new package installer
    • extracts the current versions of every dependency in the ingen_package_spec.yml from the ingen_builder conda env into ingen_package_updates.yml which mirrors ingen_package_spec.yml, but with all versions updated
    • removes the ingen_builder conda env as it is no longer needed
    • checks for updates for every dependency between ingen_package_updates.yml and ingen_package_current.yml (which contains the versions of every dependency of the currently published package installer)
    • if there are no updates of interest, the process stops, else proceeds as below
    • creates a new self extracting package installer with the steps below:
      • using the ingen_package_updates.yml, generates a set of requirements.txt and environment.yml for the new package installer
      • generates a symbiflow_installer.sh script from the template ingen_installer_script_template.sh
      • copies the conda_helper.sh, ingen_package_updates.yml and package_changelog.txt as well
      • create self-extracting package installer using makeself-2.3.1
      • naming format is: symbiflow_dailybuild_dd_MMM_YYYY.gz.run
  • [B] sanity test the generated package installer with basic tests of k6n10, k4n8, eoss3, pp3

  • [C] publish the generated package installer to Github Releases using the gh-cli + github-actions

TODO

  1. add complete CI flow automatically once the package installer is generated, tested and published
  2. optimize out the conda install, env creation if it has already been initialized once (saves time/bandwidth)
  3. document the ingen_package_spec.yml and ingen_package_current.yml formats (it is documented within already)
  4. optimize the git repo info extraction by using gh-cli rather than plain git (faster)

symbiflow-ingen's People

Contributors

coolbreeze413 avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.