Flytesnacks Contribution Guide

First off, thank you for thinking about contributing! Below you’ll find instructions that will hopefully guide you through how to contribute to and improve Flytesnacks.

💻 Contribute to Examples

  1. Determine where to put your new code

    • Core: Contains examples that demonstrate functionality available within core flytekit. These examples should be runnable locally.

    • Integrations: Contains examples that leverage one or more of the available plugins.

    • Case Studies: Contains examples that demonstrate the usage of Flyte to solve real-world problems. These are generally more complex examples that may require extra setup or that can only run on larger clusters.

  2. Create a directory (applicable for integrations and case_studies directories)

    After determining where to put your example, create a directory under the appropriate parent directory. Each example directory should contain:

    • Dockerfile

    • Makefile

    • README.rst

    • __init__.py

    • requirements.in

    • sandbox.config

    It might be easier to copy one of the existing examples and modify it to your needs.

  3. Add the example to CI

    Examples are references in this github workflow. Add a new entry under strategy -> matrix -> directory with the name of your directory as well as its relative path. Also, add the example to flyte_tests_manifest.json.

  4. Test your code!
    • If the Python code can be run locally, just use python <my file> to run it.

    • If the Python code has to be tested in a cluster:
      • Install flytectl by running brew install flyteorg/homebrew-tap/flytectl. Learn more about installation and configuration of flytectl here.

      • Run the flytectl sandbox start --source=$(pwd) command in the directory that’s one level above the directory that has Dockerfile. For example, to register house_price_prediction example, run the start command in the ml_training directory. To register core examples, run the start command in the cookbook directory. So, cd to the required directory and run all the upcoming commands in there!

        Following are the commands to run if examples in core directory are to be tested on sandbox:
        1. Build Docker container using the command: flytectl sandbox exec -- docker build . --tag "core:v1" -f core/Dockerfile.

        2. Package the examples by running pyflyte --pkgs core package --image core:v1 -f.

        3. Register the examples by running flytectl register files --archive -p flytesnacks -d development --archive flyte-package.tgz --version v1.

        4. Visit https://localhost:30081/console to view the Flyte console, which consists of the examples present in the flytesnacks/cookbook/core directory.

        5. To fetch new dependencies and rebuild the image, run flytectl sandbox exec -- docker build . --tag "core:v2" -f core/Dockerfile, pyflyte --pkgs core package --image core:v2 -f, and flytectl register files --archive -p flytesnacks -d development --archive flyte-package.tgz --version v2.

        6. Refer to Build & Deploy Your Application “Fast”er! if code in itself is updated and requirements.txt is the same.

📝 Contribute to Documentation

docs folder in flytesnacks houses the required documentation configuration. All the core, case_studies, and integrations docs are written in the respective README.md and the Python code files.

  1. README.md file needs to capture the what, why, and how
    • What is the integration about? Its features, etc.

    • Why do we need this integration? How is it going to benefit the Flyte users?

    • Showcase the uniqueness of the integration

    • How to install the plugin?

    Refer to any repo in the cookbook directory to understand this better.

  2. Explain what the code does

  3. Update conf.py (imagine you have added snowflake to the integrations/external_services folder):

    • Add the Python file names to the CUSTOM_FILE_SORT_ORDER list

    • Add ../integrations/external_services/snowflake to example_dirs

    • Add auto/integrations/external_services/snowflake to gallery_dirs

  4. Verify if the code and documentation look as expected

    1. Learn about the documentation tools here

    2. Install the requirements by running pip install -r docs-requirements.txt in the cookbook folder

    3. Run make html in the docs folder

    Tip

    For implicit targets, run make -C docs html.

    1. Open HTML pages present in the docs/_build directory in the browser

  5. After creating the pull request, check if the docs are rendered correctly by clicking on the documentation check

    Docs link in a PR