Skip to content

Latest commit

 

History

History
321 lines (191 loc) · 10.4 KB

CONTRIBUTING.rst

File metadata and controls

321 lines (191 loc) · 10.4 KB
Raw

Contributing

image

Development Installation

Additional dependencies and setup steps are required to perform a "developer installation". You do not need to perform these steps unless you intend to contribute a code or documentation change to the nucypher codebase.

Before continuing, ensure you have git installed (Git Documentation).

Acquire NuCypher Codebase

In order to contribute new code or documentation changes, you will need a local copy of the source code which is located on the NuCypher GitHub.

Note

NuCypher uses git for version control. Be sure you have it installed.

Here is the recommended procedure for acquiring the code in preparation for contributing proposed changes:

  1. Use GitHub to fork the nucypher/nucypher repository
  2. Clone your fork's repository to your local machine
$ git clone https://github.com/<YOUR-GITHUB-USERNAME>/nucypher.git
  1. Change directory to nucypher
$ cd nucypher
  1. Add nucypher/nucypher as an upstream remote
$ git remote add upstream https://github.com/nucypher/nucypher.git
  1. Update your remote tracking branches
$ git remote update

Ensure Rust is Installed

Instruction for installing Rust can be found (here).

After acquiring a local copy of the application code and installing rust, you will need to install the project dependencies, we recommend using either pip or pipenv.

Pip Development Installation

Alternately, you can install the development dependencies with pip:

$ pip3 install -e .[dev]

Running the Tests

Note

A development installation including the solidity compiler is required to run the tests

There are several test implementations in nucypher, however, the vast majority of test are written for execution with pytest. For more details see the Pytest Documentation.

To run the tests, use the following commands:

(nucypher)$ pytest -s tests/unit
(nucypher)$ pytest -s tests/integration

Optionally, to run the full, slow, verbose test suite run:

(nucypher)$ pytest

Setup Commit & Push Hooks

Pre-commit and pre-push are used for quality control to identify and prevent the inclusion of problematic code changes. They may prevent a commit that will fail if passed along to CI servers or make small formatting changes directly to source code files.

If it's not already installed in your virtual environment, install pre-commit:

(nucypher)$ pip install pre-commit

To enable pre-commit checks:

(nucypher)$ pre-commit install

To enable pre-push checks:

(nucypher)$ pre-commit install -t pre-push

For convenience, here is a one-liner to enable both:

(nucypher)$ pre-commit install && pre-commit install -t pre-push

Making a Commit

NuCypher takes pride in its commit history.

When making a commit that you intend to contribute, keep your commit descriptive and succinct. Commit messages are best written in full sentences that make an attempt to accurately describe what effect the changeset represents in the simplest form. (It takes practice!)

Imagine you are the one reviewing the code, commit-by-commit as a means of understanding the thinking behind the PRs history. Does your commit history tell an honest and accurate story?

We understand that different code authors have different development preferences, and others are first-time contributors to open source, so feel free to join our Discord and let us know how we can best support the submission of your proposed changes.

Opening a Pull Request

When considering including commits as part of a pull request into nucypher/nucypher, we highly recommend opening the pull request early, before it is finished with the mark "[WIP]" prepended to the title. We understand PRs marked "WIP" to be subject to change, history rewrites, and CI failures. Generally we will not review a WIP PR until the "[WIP]" marker has been removed from the PR title, however, this does give other contributors an opportunity to provide early feedback and assists in facilitating an iterative contribution process.

Pull Request Conflicts

As an effort to preserve authorship and a cohesive commit history, we prefer if proposed contributions are rebased over main (or appropriate branch) when a merge conflict arises, instead of making a merge commit back into the contributors fork.

Generally speaking the preferred process of doing so is with an `interactive rebase`:

Important

Be certain you do not have uncommitted changes before continuing.

  1. Update your remote tracking branches
$ git remote update
...  (some upstream changes are reported)
  1. Initiate an interactive rebase over nucypher/nucypher@main

Note

This example specifies the remote name upstream for the NuCypher organizational repository as used in the Acquire NuCypher Codebase section.

$ git rebase -i upstream/main
...  (edit & save rebase TODO list)
  1. Resolve Conflicts
$ git status
... (resolve local conflict)
$ git add path/to/resolved/conflict/file.py
$ git rebase --continue
... ( repeat as needed )
  1. Push Rebased History

After resolving all conflicts, you will need to force push to your fork's repository, since the commits are rewritten.

Warning

Force pushing will override any changes on the remote you push to, proceed with caution.

$ git push origin my-branch -f

Building Docker

Docker builds are automated as part of the publication workflow and pushed to docker cloud. However you may want to build a local version of docker for development.

We provide both a docker-compose.yml and a Dockerfile which can be used as follows:

Docker Compose:

(nucypher)$ docker-compose -f deploy/docker/docker-compose.yml build .

Release Cycle

Versioning

The versioning scheme used is inspired by semantic versioning 2.0, but adds development stage and release candidate tags. The basic idea:

  • MAJOR version when you make incompatible API changes
  • MINOR version when you add functionality in a backwards compatible manner
  • PATCH version when you make backwards compatible bug fixes

Two additional tags are used: -dev and -rc.x (i.e. v1.2.3-dev or v4.5.6-rc.0)

Upstream Branches

  • main is the stable and released version published to PyPI and docker cloud (v6.0.0).
  • development is the default upstream base branch containing new changes ahead of main and tagged with -dev (v6.1.0-dev).

Major/Minor Release Cycle

  • New pull requests are made into development.
  • When a commit from development is selected as a release candidate the version tag is changed from -dev to rc.0 (v6.1.0-rc.0). Selecting a release candidate implies a feature freeze.
  • The release candidate is deployed to beta testers, staging, and testnet environments for QA.
  • If the candidate is suitable, it is tagged, merged into main, and published:
    • All version tags are removed (v6.1.0-dev -> v6.1.0)
    • A new upstream git version tag is pushed (triggering publication on CI) (v6.1.0)
    • development is merged into main
  • development version is bumped and the -dev tag is appended (v6.2.0-dev or v7.0.0-dev)

Release Blockers

Sometimes changes are needed to fix a release blocker after a release candidate has already been selected. Normally the best course of action is to open a pull request into development.

  • Merge the pull request into development
  • Bump the release candidate's development number (v7.0.0-rc.0 -> v7.0.0-rc.1)
  • Redeploy beta testing environments, experimental nodes, staging, testnets, etc.
  • Rinse & repeat until a suitable release candidate is found.

In the event that a release blocker's fix introduces unexpected backwards incompatibility during a minor release, bump the major version instead skipping directly to -rc.0.

Patches (bugfixes, security patches, "hotfixes")

Sometimes urgent changes need to be made outside of a planned minor or major release. If the required changes are backwards compatible open a pull request into main. Once the changes are reviewed and merged, development must be rebased over main

  • Pull request is merged into main
  • The version's patch number is bumped (v6.1.0 -> v6.1.1)
  • A new upstream tag is pushed, triggering the publication build on CI (v6.1.1)
  • development is rebased over main, amending the existing bumpversion commit with the new patch (this will be a merge conflict).
  • Rinse & repeat

Release Automation

Note

This process uses towncrier and bumpversion, which can be installed by running pip install -e .[deploy] or pip install towncrier bumpversion. Also note that it requires you have git commit signing properly configured.

Important

Ensure your local tree is based on main and has no uncommitted changes.

1. Decide what part of the version to bump. The version string follows the format {major}.{minor}.{patch}-{stage}.{devnum}, so the options are major, minor, patch, stage, or devnum. We usually issue new releases increasing the patch version.

2. Use the make release script, specifying the version increment with the bump parameter. For example, for a new patch release, we would do:

(nucypher)$ make release bump=patch

3. The previous step triggers the publication webhooks. Monitor the triggered deployment build for manual approval.