Skip to content

Latest commit

 

History

History
155 lines (104 loc) · 4.36 KB

README.md

File metadata and controls

155 lines (104 loc) · 4.36 KB
Raw

Build Status Code Health Can I Use Python 3?

ep

A tool to support an explicit contract between application and plaftorm.

Inspired by Heroku's post about an explicit contract between application and platform, ep aims to provide a reusable mechanism to cover that contract.

As a summary, Heroku's contract is based on:

  • Dependency management: declared as part of the codebase
  • Procfile: as a mechanism to specify what should be run
  • Web process binds to $PORT (an externally provided environment variable
  • stdout for logs
  • Resource handles in the environment: environment variables for configuration, in particular for connecting to external resources

What defines a contract in ep

An ep-based contract tries to cover these areas:

  • A defined way to specify your dependencies
  • A defined way to run your project
  • Configured purely via environment variables

The particulars of the contract are specified via a configuration file that lives on the root of your application's repository. This configuration file is in YAML format and is called ep.yml by default.

A defined way to specify dependencies

Dependencies can be of various types. ep aims to cover at least the following:

  • python: via pip install of a requirements file. Defaults to requirements.txt at the root folder
  • node: via npm install of a package file. Defaults to package.json at the root

The root here refers to the directory where ep.yml lives.

Future extensions may cover ruby gems, bower packages, and even os-level packages on some OSes

There are two types of dependencies from the PoV of ep:

  • Those that the tool will be able to install
  • Those that it will only verify, but not attempt to install. These include e.g. the version of Python in your system

Some of these dependencies will have their defaults. This could be a partial example of a ep.yml.

ep: 1.0.0
dependencies:
  - python
      version: 2.7
      file: requirements/runtime.txt
  - npm

The command to install dependencies via ep is:

ep setup

ep will always install dependencies in an isolated environment. This means e.g. a virtual environment in Python, local npm install etc. ep will manage the creation of this isolated environment for you.

You can always delete installed dependencies for a fresh initial state:

ep clear

A defined way to run your project

ep doesn't force you to use Procfile, but lets you define your run command:

run: gunicorn myapp.wsgi

Defaults to running a Procfile in your root.

Run your project using:

ep run

You can also run multiple commands:

run:
  - echo "Starting ..."
  - gunicorn myapp.wsgi

Configuration via environment variables

Your app should be configurable via environment variables. Your ep.yml file provides an explicit definition of what those variables are, and optional default values and help.

You should include in an env section the environment variables your system relies on for running, add help text for them, and when applicable define some defaults:

env:
  PORT:
    help: "The port the web application will run on"
  SOME_EXTERNAL_SERVICE_URL:
    help: "URL to your external service blah."
    default: "http://localhost:9000"

ep run will complain if variables that do not have a default value are not provided, and refuse to run.

Performing additional checks

Besides default checks for dependency management and environment variables, you can add custom check steps. As for run, these can be a single string or a list:

check:
  - python -m unittest discover
  - flake8 ep

Additional features

Besides the basic explicit contract defined above, ep helps you with extra things.

Some commands

ep shell runs commands within ep's isolated environment. E.g.:

ep shell "pip list"

Extra setup steps

TODO: define a mechanism to have additional setup steps such as running migrations on a django project, etc...