Skip to content

Latest commit

 

History

History
505 lines (355 loc) · 30.9 KB

COLLABORATOR_GUIDE.md

File metadata and controls

505 lines (355 loc) · 30.9 KB
Raw

Node.js Collaborator Guide

This document contains information for Collaborators of the Node.js website project regarding maintaining the code, documentation, and issues.

Collaborators should be familiar with the guidelines for new contributors in CONTRIBUTING.md.

Issues and Pull Requests

Courtesy should always be shown to individuals submitting issues and pull requests to the Node.js website project.

Collaborators should feel free to take full responsibility for managing issues and pull requests they feel qualified to handle, as long as this is done while being mindful of these guidelines, the opinions of other Collaborators and guidance of the Website Group.

Collaborators may close any issue or pull request they believe is not relevant to the future of the Node.js project. Where this is unclear, the issue should be left open for several days for additional discussion. Where this does not yield input from Node.js Collaborators or additional evidence that the issue has relevance, then the issue may be closed. Remember that issues can always be re-opened if necessary.

Important

We recommend Collaborators to avoid Updating/Rebasing PRs unnecessarily, since we use GitHub Merge Queues to merge Pull Requests, which automatically rebases and runs CI-checks against the latest base branch.

Accepting Modifications

All Node.js code and documentation modifications should be performed via GitHub pull requests. Only the Website Team can merge their work and should do so carefully.

All pull requests must be reviewed and accepted by a Collaborator with sufficient expertise who can take full responsibility for the change. In the case of pull requests proposed by an existing Collaborator, an additional Collaborator is required for sign-off.

Pull Requests can only be merged after all CI Checks have passed. As usual, CI Checks need to be manually triggered by adding a github_actions:pull-request label to the Pull Request.

In some cases, it may be necessary to summon a qualified Collaborator to a pull request for review by @-mention.

If you are unsure about the modification and are not prepared to take full responsibility for the change, defer to another Collaborator.

We recommend collaborators follow the guidelines on the Contributing Guide for reviewing and merging Pull Requests.

Involving the Website Team

Collaborators may opt to elevate pull requests or issues to the group for discussion by mentioning @nodejs/nodejs-website. This should be done where a pull request:

  • has a significant impact on the codebase,
  • is inherently controversial; or
  • has failed to reach a consensus amongst the Collaborators who are actively participating in the discussion.

The Website group should be the final arbiter where needed.

Technologies used in the Website

The Node.js Website is built upon React and Next.js respectively, the UI Rendering Engine and the Framework that builds the Website;

The Website also uses several other Open Source libraries (not limited to) listed below:

  • Styling is done with PostCSS and CSS Modules
  • Tailwind is used as our CSS Framework and the Foundation of our Design System
  • Hero Icons is an SVG Icon Library used within our Codebase
  • Radix UI is a collection of customizable UI components
  • Shiki is a Syntax Highlighter used for our Codeboxes
    • The syntax highlighting is done within the processing of the Markdown files with the MDX compiler as a Rehype plugin.
  • MDX and Markdown are used for structuring the Content of the Website
  • next-intl is the i18n Library adopted within the Website
    • It provides an excellent integration with Next.js, But it also supports standalone support for i18n if it eventually migrates from Next.js to something else.
    • Supports React Server Components and Next.js Middlewares
  • next-sitemap is used for Sitemap and robots.txt Generation
  • We use Rehype and Remark to extend MDX functionality
  • We use Storybook for Manual Testing and Visual Regression Tests of our React Components
    • Storybook also provides a sandboxed environment, which is very useful whilst for crafting React Components
  • We use Sentry for reporting Exceptions and monitoring the performance and reliability of the application

Code Editing

Structure of this Repository

⚠️ The repository is actively under migration to a multi-package workspace. Locations are subject to change. (If you are someone updating these paths, please document those changes here.)

  • React Components are defined on apps/site/components
  • React Templates are defined on apps/site/layouts
  • Global Stylesheets are declared on apps/site/styles
  • Public files are stored on apps/site/public
    • Static Images, JavaScript files, and others are stored within apps/site/public/static
  • Internationalisation is done on apps/site/i18n
    • React Localisation Data is stored within apps/site/i18n/locales
    • We use the ICU Message Syntax for Translations
    • Configuration for Locales is done within apps/site/i18n/config.json
  • Website Content is defined within apps/site/pages
    • Initial development usually happens in English: apps/site/pages/en
    • Localized versions of /pages/en are done within apps/site/pages/{localeCode}
    • All content is in Markdown and is per locale.
    • The top of each Markdown file is a YAML (Frontmatter) block for page-specific localization information passed to various templates.
    • The bulk of the Markdown content for each page is referenced as {children} on their respective JSX Layout (apps/site/layouts/)
  • Multi-Purpose React Hooks are defined on apps/site/hooks
  • Multi-Purpose TypeScript definitions are defined on apps/site/types
  • React Context Providers are defined on apps/site/providers
  • Build-time Data Fetching Scripts are defined on apps/site/next-data
    • Used for Node.js Release data fetching
    • Generation of build-time indexes such as blog data
  • Multi-Purpose Scripts are stored within apps/site/scripts
    • Such as Node.js Release Blog Post generation
  • Storybook Configuration is done within apps/site/.storybook
    • We use an almost out-of-the-box Storybook Experience with a few extra customisations

Adding new Pages

  1. Create new page content including the layout, title and copy.
  2. Update the relevant apps/site/layout to add a link to the new page.

Create the page content

Create a new markdown file in apps/site/pages/en.

At the top of the markdown file, within the Markdown Frontmatter, set a page the title and layout.

---
title: Title of the Page
layout: layout-name
---

[Content of the Page]

Note

A list of currently available Layouts is provided within components/withLayout on the layoutComponents map.
This is a temporary map and this map might change its location and be defined in a different way in the future.

Translating Pages

See the Translation Guidelines for the website translation policy.

Creating React Components

The Node.js Website uses React as a Frontend Library to develop the Website. React allows us to create user interfaces with a modern take on Web Development.

If you're unfamiliar with React or Web Development in general, we encourage a read before taking on complex issues and tasks as this repository is not for educational purposes and we expect you to have a basic understanding of the technologies used.

We also recommend getting familiar with technologies such as Next.js, MDX, PostCSS, and "concepts" such as "CSS Modules" and "CSS-in-JS".

Styling a Component

As mentioned, we write all Component-styles in separate .module.css files. This is like writing any CSS in a separate file (besides the fact that we use PostCSS).

This concept of writing styles on dedicated CSS files and importing them within JavaScript (or React) is a pattern named CSS Module. These allow us to write PostCSS (or regular CSS, or any flavor of CSS if you have a way of interpreting it) within a .module.css and import the class names directly to our React Components. We recommend reading guides on "Styling React Components with CSS Modules", which there are many available on the web.

It's important to mention that we use Tailwind as a CSS Framework. Hence, margins, paddings, font sizes, font weights, colors, and other sorts of styles are all provided with Tailwind. We recommend reading Tailwind Docs to get familiar with Tailwind's styles. We also recommend reading this guide for setting up Tailwind on your IDE.

Finally, if you're unfamiliar with how to use Tailwind or how to use Tailwind within CSS Modules, we recommend reading this guide.

Example of a CSS Module

.myComponent {
  @apply some
    tailwind
    classes;
}

Guidelines when writing CSS

  • We use camelCase for defining CSS classes
  • We use Tailwind's @apply selector to apply Tailwind Tokens
    • We discourage the usage of any plain CSS styles and tokens, when in doubt ask for help
    • We require that you define one Tailwind Token per line, just as shown on the example above, since this improves readability
  • Only write CSS within CSS Modules, avoid writing CSS within JavaScript files
  • We recommend creating mixins for reusable animations, effects and more
    • You can create Mixins within the apps/site/styles/mixins folder

Note

Tailwind is already configured for this repository. You don't need to import any Tailwind module within your CSS module.
You can apply Tailwind Tokens with Tailwind's @apply CSS rule. Read more about applying Tailwind classes with @apply.

Important

When using IDEs such as Visual Studio Code, we recommend installing the official Stylelint and Tailwind Extensions.
These are recommended Extensions for IntelliSense, Syntax Highlighting and Error Checking when styling your Component.

Best practices when creating a Component

  • All React Components should be placed within the apps/site/components folder.
  • Each Component should be placed, whenever possible, within a sub-folder, which we call the "Domain" of the Component
    • The domain represents where these Components belong to or where they will be used.
    • For example, Components used within Article Pages or that are part of the structure of an Article or the Article Layouts, should be placed within apps/site/components/Article
  • Each component should have its folder with the name of the Component
  • The structure of each component folder follows the following template: 
    - ComponentName
  - index.tsx // the component itself
  - index.module.css // all styles of the component are placed there
  - index.stories.tsx // component Storybook stories
  - __tests__ // component tests (such as unit tests, etc)
    - index.test.mjs // unit tests should be done in ESM and not TypeScript
  • React Hooks belonging to a single Component should be placed within the Component's folder
    • If the Hook as a wider usability or can be used by other components, it should be placed in the root hooks folder.
  • If the Component has "sub-components" they should follow the same philosophy as the Component itself.
    • For example, if the Component ComponentName has a sub-component called SubComponentName, then it should be placed within ComponentName/SubComponentName

How a new Component should look like when freshly created

import type { FC } from 'react';

import styles from './index.module.css';

type MyComponentProps = {}; // The types of the Props of your Component

const MyComponent: FC<MyComponentProps> = ({ prop1, prop2... }) => (
  // Actual code of my Component
);

export default MyComponent;

Best practices for Component development in general

  • Only spread props { ... } on the definition of the Component (Avoid having a variable named props)
  • Avoid importing React, only import the modules from React that you need
  • When importing types, use import type { NameOfImport } from 'module'
  • When defining a Component, use the FC type from React to define the type of the Component
    • When using children as a prop, use the FC<PropsWithChildren<MyComponentProps>> type instead
    • Alternatively you can define your type as type MyComponentProps = PropsWithChildren<{ my other props }>
  • Each Props type should be prefixed by the name of the Component
  • Components should always be the default export of a React Component file
  • Avoid using DOM/Web APIs/document/window API access within a React Component. Use utilities or Hooks when you need a Reactive state
  • Avoid making your Component too big. Deconstruct it into smaller Components/Hooks whenever possible

Unit Tests and Storybooks

Each new feature or bug fix should be accompanied by a unit test (when deemed valuable). We use Jest as our test runner and React Testing Library for our React unit tests.

We also use Storybook to document our components. Each component should have a storybook story that documents the component's usage.

Visual Regression Testing is automatically done via Chromatic to ensure that Components are rendered correctly.

General Guidelines for Unit Tests

Unit Tests are fundamental to ensure that code changes do not disrupt the functionalities of the Node.js Website:

  • Unit tests should be written as .mjs files.
  • We recommend adding unit tests for content covering util, scripts, hooks, and components whenever possible.
  • Unit Tests should ensure that a given change's functionality is working as expected.
  • When creating unit tests for React components, we recommend that the tests cover all the possible states of the component.
  • We also recommend mocking external dependencies, if unsure about how to mock a particular dependency, raise the question on your Pull Request.
    • We recommend using Jest's Mock Functions for mocking dependencies.
    • We recommend using Jest's Mock Modules for mocking dependencies unavailable on the Node.js runtime.
    • Common Providers and Contexts from the lifecycle of our App, such as next-intl should not be mocked but given an empty or fake context whenever possible.
  • We recommend reading previous unit tests from the codebase for inspiration and code guidelines.

General Guidelines for Storybooks

Storybooks are an essential part of our development process. They help us to document our components and to ensure that the components are working as expected.

They also allow Developers to preview Components and be able to test them manually/individually to the smallest unit of the Application. (The individual Component itself).

Storybooks should be fully typed and follow the following template:

import type { Meta as MetaObj, StoryObj } from '@storybook/react';
import NameOfComponent from '@components/PathTo/YourComponent';

type Story = StoryObj<typeof NameOfComponent>;
type Meta = MetaObj<typeof NameOfComponent>;

// If the component has any props that are interactable, they should be passed here
// We recommend reading Storybook docs for args: https://storybook.js.org/docs/react/writing-stories/args
export const Default: Story = {};

// If the Component has more than one State/Layout/Variant, there should be one Story for each variant
export const AnotherStory: Story = {
  args: {},
};

export default { component: NameOfComponent } as Meta;
  • Stories should have args whenever possible, we want to be able to test the different aspects of a Component
  • Please follow the template above to keep the Storybooks as consistent as possible
  • We recommend reading previous Storybooks from the codebase for inspiration and code guidelines.
  • If you need to decorate/wrap your Component/Story with a Container/Provider, please use Storybook Decorators

Remarks on Technologies Used

The Node.js Website is a somewhat complex application and at times non-trivial solutions have been implemented to solve certain technical challenges. Historical decision making can be largely found on past issues, conversations on Slack and GitHub discussions. However, we also wish to highlight some of the notable development decisions that we have made here.

Why Next.js?

We've found that Next.js is simply versatile, hackable, stable, community-maintained and has a great ecosystem. The reasoning goes deeper, but as a long-term Framework it is the most suitable choice.

Why do we continue to support static builds?

It was decided together with the TSC (Technical Steering Committee) that the Node.js Website should always support fully static builds that do not depend on any 3rd party services. This is to ensure that the Website is always available and that we do not depend on any 3rd party services to serve our content.

(For example, if we abandon Vercel, our Website should still completely work as standalone as possible)

What is next.dynamic.mjs?

Our whole Website uses a custom renderer for rendering the pages. As you might have seen, within the apps/site/pages directory we have Next.js Dynamic Route named [...path].tsx that matches against all possible routes of the Website.

This means that each .md(x) file within apps/site/pages/ is not rendered by Next.js regular App Tree (apps/site/pages/_document.tsx and apps/site/pages/_app.tsx) but a custom render tree.

This custom render uses getStaticPaths and Incremental Static Generation to generate the full list of supported pages of the Website. For example, this allows us to generate Localized Pages for every page that is not translated, by telling Next.js to create a localised path. next.dynamic.mjs is responsible for getting a full list of the source pages (apps/site/pages/en) and identifying which pages have been translated.

Non-translated pages will have their Localized contexts and translated React message-bags (next-intl) but the content will be the same as the source page (English). Whereas localized pages will have localized context and content.

This custom solution is also able to decide what paths should be compiled during runtime. This is a combination of rules defined on next.constants.mjs and [...path].tsx.

The [...path].tsx file ultimately utilizes the theme.tsx file as its layout source. This setup enables the loading of the Layout Provider and MDX Provider, which in turn, encapsulate and manage the rendering of any child content or components. This includes both content and components provided by the Layout Provider and the transformed MDX content originating from the .md(x) source page.

What is next.data.mjs?

This file is responsible for loading, fetching and generating build-time required information such as Node.js Release data, Blog Posts Metadata (for pagination and indexation), RSS feeds, etc.

What is site.json?

This file is used for defining Website Metadata, such as which RSS feeds should be generated, Social Media Information and other Metadata used during the Website build-time. We use a JSON format to ease collaboration.

What is next.locales.mjs and why not use Next.js built-in i18n?

While Next.js provides a built-in i18n feature, it doesn't offer the flexibility we require. Our specific needs include the ability to generate comprehensive lists of supported locales and pages for each locale. Additionally, we operate with a subfolder-based approach for content, as opposed to the extension-based approach (e.g., filename.language.md(x)) that is compatible with Next.js's built-in i18n.

We opted for the subfolder approach to maintain consistency with our previous Node.js website's content structure and to ensure long-term maintainability, rather than relying on Next.js's i18n functionality.

What is next.rewrites.mjs?

This file is responsible for defining the rewrite rules for the Website. It is used for defining redirects and other rewrite rules. (Such as Internal Redirects and External ones).

The redirects.json file specifies two types of operations: rewrites and redirects.

  • Rewrites: These operations do not alter the original URL but instead render the content from the rewritten URL. It's important to note that the rewritten URL path must be valid and exist on the website.
  • Redirects: Redirect operations, on the other hand, change the original URL to a new one. While these new URLs can be either internal (within the website) or external (leading to a different domain), it is more common for redirects to point externally.

This file contains a simple template engine that supports /:locale to indicate that this path should also be available under all available locales as prefix.

Why do we use Next.js Middlewares?

We have a simple Next.js Middleware that is responsible for handling initial Locale detection and redirection. It detects browser locales and redirects to the most suitable locale for the user. And it fallbacks to /en if no suitable locale is found.

What are Layouts?

Layouts Wrap the content of the Markdown files. They are responsible for adding additional styling and structure surrounding the content of the Markdown files.

Layouts are defined within the layouts folder. They are React Components that receive the children prop, which is the transformed MDX content of the Markdown file.

Each Page layout is configured within their Markdown's Frontmatter as layout: name-of-layout.

How we style the Website?

We use PostCSS to style the Node.js Website; PostCSS is a CSS Preprocessor, like Sass and Less.

How exactly do we style Components?

We style each individual React Component with a dedicated CSS file (A CSS Module) that uses CSS syntax (with the extra powerups of PostCSS).

The Styling a Component section contains a more detailed guide on how we style our Components.

Why we use PostCSS over Sass or Less?

The main advantage of PostCSS is its minimal pluggable API that allows us to extend the native CSS-syntax with custom plugins.

Next.js natively supports PostCSS and always uses PostCSS as part of the bundling and building process. By not using Sass or Less we remove another dependency from our build process and remove one layer of preprocessing our styles (CSS).

We currently use a set of PostCSS plugins that create a SCSS-a-like environment.

What PostCSS Plugins we use?

  • postcss-mixins: Allows us to use Sass-like Mixins
  • postcss-import: Allows us to use Sass-like Imports
  • postcss-simple-vars: Allows us to use Sass-like Variables
  • postcss-nested: Allows us to use Sass-like Nesting
  • postcss-calc: Strips calc expressions and replaces them with the result

It is important to mention that even though we use SCSS-like syntax, we do not use SCSS, and some of these plugins are not 100% compatible with the SCSS syntax. For example, postcss-mixins does not support @include and @extend directives (and it uses @define-mixin for defining Mixins and @mixin for including Mixins).

Do we use a CSS Framework?

The Node.js Website uses Tailwind as a CSS Framework for crafting our React Components and style the Website.

Tailwind is an utility-first CSS Framework. It allows us to create a Design System that is easy to maintain and extend. It also allows us to create a consistent Design Language across the Website.

Font Families on the Website

We use next/fonts Open Sans as the default font for the Node.js Website. The font is configured as a CSS variable and then configured on tailwind.config.js as the default font for the Website.

Why we use RadixUI?

  • It is a minimalistic component library broken down in individual packages for each Component
  • It already handles all WAI-ARIA and Accessibility shortcuts/bindings needed for Interactive Elements
  • It allows us to focus on designing interactive Components without the effort of adding all the surrounding sugar and code needed to make the Component accessibility-friendly.

Why MDX?

MDX is an extension on Markdown that allows us to add JSX Components within Markdown. Besides that, MDX is also a pluggable parser built on top of unified which supports Rehype and Remark Plugins. MDX is becoming the standard for parsing human-content on React/Next.js-based Applications.

Some of the plugins that we use include:

  • remark-gfm: Allows us to bring GitHub Flavoured Markdown within MDX
  • remark-headings: Generates Metadata for Markdown Headings
    • This allows us to build the Table of Contents for each Page, for example.
  • rehype-autolink-headings: Allows us to add Anchor Links to Markdown Headings
  • rehype-slug: Allows us to add IDs to Markdown Headings

Syntax Highlighting (Shiki) and Vercel

Shiki is integrated on our workflow as a Rehype Plugin, see the next.mdx.shiki.mjs file. We also use the nord theme for Shiki and a subset of the supported languages as defined on the shiki.config.mjs file.

Vercel

We use Vercel as our hosting provider. It is a great platform that offers an excellent CI/CD pipeline which allows us to deploy our website with ease.

It is important to mention that there are some rules on our Vercel Deployments such as:

  • Branches starting with dependabot (Dependabot Automated PRs) or gh (GitHub Merge Queues) are not deployed to Vercel.
  • Vercel Deployments are triggered for all other branches during push activities.
  • We have a custom install script that executes npm ci --omit=dev (the same way we do on our CI Pipelines)
    • Hence if Builds fail unexpectedly, make sure that your dependency that is being used during build-time is on dependencies and not devDependencies. Checkout out DEPENDENCY_PINNING.md for more information.
  • Our sponsorship with Vercel is maintained by the OpenJS Foundation

Why we have a .vscode folder

The repository defines an optimized configuration for code editing. This is optional and is not required to contribute to the project. However, the settings and extensions specified help create a uniform and more efficient developer experience. This configuration is found in the .vscode directory:

  • extensions.json suggests VSCode extensions that make the editor more compatible with the code. For example, a Tailwind extension creates auto-complete intellisense for tailwind styles within our components. Eslint, prettier, and editorconfig extensions read their respective config files and automatically format or lint code as written. This helps save CI feedback loops when a contribution does not meet our standards.
  • settings.json contains some common sense defaults that aide development and enforce consistency across the codebase. For example, we want files formatted on save and we want prettier to be used as the formatter. Without these settings, new contributors may have different authoring experiences when contributing, leading to inconsistent code and CI failures. We also disable VSCode's default CSS parser so PostCSS and Tailwind syntax are respected.

Defining a .vscode configuration like this also aides browser-only development using GitHub's Codespaces feature. The web-based GUI will read these same configuration files and setup the remote development environment the same way every time.

Why we have an .npmrc file

The npm ecosystem resolution and installation of peerDependencies installation changed in recent versions. The project documents what version of Node.js and npm to use via the .nvmrc file. Not all contributors have tooling to automatically read this file and adhere to the correct version, however. To ensure all contributors install dependencies the same way, a local .npmrc file directly configures peerDependency installation behavior.

Seeking additional clarification

A lot of the current structure is due to retro-compatibility, keeping a simple and familiar file structure and keeping files that have historical reasons or needs.

If you're unfamiliar or curious about something, we recommend opening a Discussion on this GitHub Repository.