Perry is scaffolded using tsdx.
TSDX scaffolds your new library inside /src.
To run TSDX, use:
yarn startThis builds to /dist and runs the project in watch mode so any edits you save inside src causes a rebuild to /dist.
To do a one-off build, use yarn build.
To run tests, use yarn test.
Code quality is set up for you with prettier, husky, and lint-staged.
Jest tests are set up to run with yarn test.
size-limit is set up to calculate the real cost of your library with yarn size and visualize the bundle with yarn analyze.
TSDX uses Rollup as a bundler and generates multiple rollup configs for various module formats and build settings. See Optimizations for details.
tsconfig.json is set up to interpret dom and esnext types, as well as react for jsx. Adjust according to your needs.
Two actions are added by default:
mainwhich installs deps w/ cache, lints, tests, and builds on all pushes against a Node and OS matrixsizewhich comments cost comparison of your library on every pull request usingsize-limit
Please see the main tsdx optimizations docs. In particular, know that you can take advantage of development-only optimizations:
// ./types/index.d.ts
declare var __DEV__: boolean;
// inside your code...
if (__DEV__) {
console.log('foo');
}You can also choose to install and use invariant and warning functions.
CJS, ESModules, and UMD module formats are supported.
The appropriate paths are configured in package.json and dist/index.js accordingly. Please report if any issues are found.
Per Palmer Group guidelines, always use named exports. Code split inside your React app instead of your React library.
Run yarn version. Maintainers with publishing rights alone shall be able to publish to NPM.
- Use the following structure for naming any feature branch:
${username}/#${issue-number}-${some-title}. - You can read more about the reasoning here.
- The first portion of the branch name
${username}/...is a "grouping token". It helps in clubbing all the branches owned by a particular user.
We follow the Git Feature Branch Workflow for pushing new features into the master branch.
- We use the conventional commits specification to decide how to format our commit messages.
- This is not strictly enforced at the moment but the developers are expected to follow this specification while commiting important changes like bringing in a new feature, making a breaking change to the API, or any change that the developer feels should be reflected in the CHANGELOG in the subsequent release.
- Conventional commits integrate well with the SemVer versioning pattern and this compatibility is the main reason for its adoption.
conventional-changelog-cliis used in theyarn versionCLI giving us the power to automate the CHANGELOG generation.- The Conventional Commits extension makes the formating & compliance of the commit messages a cake-walk!
