Engineering Onboarding Guide

Code Style

Some developers feel the need to debate code style. Other engineers will use the project's style and get stuff done. We prefer engineers who get stuff done. Consistency is important, please follow these guidelines.

We use the JavaScript Standard Style in our code. This is the style used by npm, GitHub, Zeit, MongoDB, Express, Electron, and many others. Be sure to use an automatic formatter like standard-formatter for Atom and vscode-standardjs for Visual Studio Code.

Line length should be limited to 80 characters.

Code Principles

• Code should be as simple, explicit, and as easy to understand as possible.
• Functional style is preferred to OOP. When possible functions should be pure and not rely on shared state or side effects.
• Avoid large frameworks; use small modules that are easy to understand.
• Modules must use this order so that they can be understood quickly when skimmed:
  1. External dependencies: anything listed in package.json, e.g. require('http')
  2. Internal dependencies: any files created in the project itself, e.g. require('./api')
  3. Constants and other setup: this includes anything absolutely necessary to be defined before module.exports
  4. Exports: module.exports should be as close to the beginning of the file as possible. The module should export either a single function or a "catalog object", e.g. module.exports = { method1, method2, ... }
  5. Functions: these go after the above sections. Use function hoisting to control the placement of your functions so that important, high-level functions are above smaller more-general utility functions.
• Use descriptive variable names. Function names should be a verb like route() or verb combined with a noun like routeRequest().
• Keep your functions short. If your function is over 40 lines, you should have a good reason.
• Functions should not accept more than 3 arguments. Use a single options object if you need more arguments.
• Keep nesting to a minimum. Use early returns, single-line conditionals, and function calls.

Git & Github

Version control is a project's best source of documentation when done correctly. When trying to understand code it's extremely useful to use git blame to find both the PR and the issue associated with that change.

PRs should be small and focused. Each commit should solve a single problem and be covered by a test that exemplifies that particular feature or fix.

All PRs must be reviewed by a teammate before they are eligle to be merged into the master branch. Large PRs are difficult to review. Be sure to break large PRs into smaller ones so that they can be reviewed quickly and deployed to production.

• Never commit passwords, access tokens, or other credentials into version control. If you think you absolutely have to, ask first. If you do this by accident, tell someone immediately.
• Each commit should be as small and as simple as possible.
• The project must be operational and have all tests passing after every commit.
• Use Conventional Commits
  • See the "Commit Types" section below
  • Valid types are chore:, docs:, style:, refactor:, perf:, and test:
• Do not mix feature changes (added functionality) with fixes (restored functionality), refactors (no change in functionality), or style changes (only whitespace or other cosmetic changes).
• Before a PR is ready for review, make sure that it is a single commit. If the combined commit is too large or disparate, consider multiple PRs.
• The exception to the above single commit rule is when a PR introduces new packages. Create one extra commit in the same PR for each new package your PR needs.
• Do not modify a project's .gitignore to add files related to your editor or environment. Use your own global .gitignore for that instead.
• Be sure that your PRs have descriptive titles that explain what has been changed. Typically the commit message is sufficient. "Fixes #66" is not.
• When submitting a PR with UI or visual changes, please add before and after screenshots to the PR. This makes it easy for the reviewer to quickly see what has been done.

Workflow

We use Zenhub to manage our workflow. Each task is represented by an issue. Be sure to connect any PR you are working on to the appropriate issue. Do this via the Zenhub interface and by adding Closes #X where X is the issue number to the PR description in Github.

As tasks move from Backlog to In Progress to Needs Review to Needs QA to Ready to Deploy and finally to Closed.

• Backlog: This is where you will choose from issues to work on. Once you have selected one, move it to In Progress.
• In Progress: While you are working on an issue, it should stay in this column. Once you are finished and satisfied with your work, move it to Needs Review.
• Needs Review: Each day you should be looking at this column for PRs to review. This column will list all issues that are complete and are waiting on review before they can be ready for deploy. When you review an item, make sure that it follows all of our coding principles and will not cause problems when we deploy it to production. Once a PR has been reviewed, either move it back to In Progress or forward to Needs QA or Ready to Deploy. Do not allow items to sit in Needs Review.
• Needs QA: If the change affects the front-end and can be tested with a deploy preview it should be tested by QA. After successful testing it should be moved to Ready to Deploy, but if it is unsatisfactory, it should be moved back to In Progress.
• Ready To Deploy: After an issue is finished and has been reviewed and approved by a teammate, it will be moved to this column. After it has been deployed, it will be moved to the Closed column.

Slack & Pull Requests

We have a dedicated Slack channel for posting PRs: #engineering-prs. Every PR should be posted in this channel, allowing for others to know that you have work that needs review. Communicating updates about the PR will happen in a thread attached to the original Slack message for the PR.

The process for posting a PR in Slack is:

• Post your PR in #engineering-prs
• If you are the developer reviewing the PR, start a thread, saying: @<developer_who_made_PR> reviewing
• If the PR needs to be pushed back, the developer reviewing responds in the thread with: pushing back to In Progress @<developer_who_made_PR>.
• If all comments have been addressed, the developer responds in the thread with: @<developer_who_made_the_comment> all comments have been addressed. Pushing back to Needs Review
• If you have approved the PR, respond in the thread with: @<developer_who_made_PR> approved

Please note that all discussion about the PR should stay in GitHub. Threads are not intended a place to discuss the code itself. Slack is only used to provide more immediate feedback.

Commit Types

Commit types (e.g. feat, fix, refactor, style) are important because they are a signal to the reviewer for what they should be looking for and how much scrutiny the review needs. Reference from Angular:

• feat: A new feature
• fix: A bug fix
• refactor: A code change that neither fixes a bug nor adds a feature
• style: Changes that do not affect the meaning of the code (white-space, formatting, missing semi-colons, etc)
• test: Adding missing tests or correcting existing tests
• docs: Documentation only changes
• perf: A code change that improves performance
• build: Changes that affect the build system or external dependencies (example scopes: gulp, broccoli, npm)
• ci: Changes to our CI configuration files and scripts (example scopes: Travis, Circle, BrowserStack, SauceLabs)

Descending Levels of Scrutiny and Test Requirements

Feature commits require the most scrutiny. They add or change functionality, require additional tests, and have the greatest chance of introducing a bug.

Fix commits are smaller and have a more bounded scope than feature commits. Features usually introduce multiple new behaviors, but fix commits will only modify a single behavior. Fix commits need to provide additional test coverage because the existing tests were insufficient to catch the error. Feature commits should have multiple tests for these behaviors, but fix commits typically introduce a single test to prove that the error has been fixed (test should fail without the fix inactive and pass with it active). These commits can introduce additional bugs, but it’s less common than feature commits.

Refactors need less scrutiny than feature or fix commits because little about the app changes. All consumers of the refactored part of the app should be 100% unaware of the change. If module A is refactored, and module B depends on A, module B does not need to be tested because A hasn’t changed anything from their perspective. Similarly, anything that depends on B doesn’t need to be tested either, no change should bubble up to cause issues. Since nothing is changing, any existing tests/QA should already be sufficient to catch any issues created by the refactor. Refactors don’t generate additional tests.

Style commits need the least amount of scrutiny. They tend to be very repetive (converting from snake_case to camelCase, standardizing indentation, or changing newlines). These changes will have no functional impact on the code and any existing tests should be sufficient.

Slack and Communication

All communication about projects should be in #engineering. Do not use direct messages unless you are discussing something private. It is important that all conversations and questions (no matter how small they seem) should be in #engineering. We work as a team and that requires sharing, being transparent, and allowing the whole team to have a chance to answer your questions and to learn from the dialogue.