choo-persist

choo-persist

Synchronize choo's state with indexedDB. Does nothing in browsers that don't support indexedDB.

Usage

const persist = require('choo-persist')
const choo = require('choo')

const app = choo()

/* register router, views, models */

persist((persist) => {
  app.use(persist)
  const tree = app.start()
  document.body.appendChild(tree)
})

API

persist(opts?, fn(plugin))

Create a new indexedDB database instance, and call the callback with the plugin when done. Can take an optional first argument of options:

• opts.name: default 'app'; provide a name for the indexedDB database

plugin.onStateChange

The plugin should be passed into app.use() directly. onStateChange listens for every state change in the application and persists it to indexedDB.

plugin.wrapInitialState

The plugin should be passed into app.use() directly. wrapInitialState reads the state from indexedDB on start and overrides the initial application state.

Installation

$ npm install choo-persist

FAQ

Why does this plugin wrap the start() method?

Choo is initialized completely synchronously. This makes it easy to reason about, implement, and profile. IndexedDB, however, cannot be accessed synchronously so in order to get the initial state from the DB we must wait for it to load first. This is why the application start must be wrapped in this plugin.

Why not load the default initial state first, and then apply indexedDB?

If the default initial state is loaded from the application before indexedDB kicks in, there will be a flash of unstyled content. It's better to have generic server side rendering (e.g. can be cached by CDNs) to lower the time to first rendering, and then load the initial state from the local database after that to serve the first uncacheable content. This will provide a smooth experience; and with the addition of a few CSS transitions it might become impressive even.

Why would I want to load content from indexedDB?

Because time to first render matters, and connectivity might not always be great, or available even. By synchronizing the application state with indexedDB the biggest win for offline has already been achieved: users will be able to pick up right where they left off; even if internet goes down. From there on out, caching outgoing HTTP requests and other fancy connectivity tricks could be applied. Consider this package as an easy way to get a big win for offline first.

Should I use this while developing.

No; state is persisted between page reloads which might put your page in very odd states, with a very annoying way to clear. Consider using hot reloading for development instead.

How / when should I invalidate the database cache?

Not sure yet; if you've got good ideas for upgrading / invalidating the local database I'd love to hear from you; I'm sure people will run into this down the road anyway.

Can I use this with something other than choo?

Welp, there's not much code but if it talks the same API then yeah sure you can. This is a bit specific to choo tho; but there's nothing stopping people from building their own choo soooo...

License

MIT