intern-resources/readme.md

thunderstrike.ai intern resources

purescript

Purescript is a strongly-typed ^{[type safety]} purely functional ^{[purely functional]} programming language that compiles to javascript (like typescript).

In brief

• effectful code ^{[side effect]} is isolated from non-effectful code. This makes programs much more consistent to run and reason about.
• everything is an expression
• interacting with raw javascript is relatively painless

Motivating example: React vs Halogen

this frontend app:

1. renders "Loading..." until either data or error
2. fetches data from a server
3. if ok, render items in a list. if error, render error span at the bottom of the page.

React

Halogen

export const App = () => { const [items, setItems] = React.useState(undefined) const [error, setError] = React.useState(undefined) React.useEffect(() => { fetch('/api/item') .then(rep => rep.json()) .then(items => setItems(items)) .catch(e => setError(e)) }, []) return error !== undefined ? (<p class="error">{error.message}</p>) : items === undefined ? (<p>Loading...</p>) : items.map(item => { return ( <div key={item.id}> <p>{item.title}</p> <span class="small"> {item.description} </span> </div> ) }) }

type Item = {id :: String, title :: String, description :: String} data State = StateEmpty | StateErrored Error | StateGotItems (Array Item) data Action = Init renderItem item = div [] [ p [] [text item.title] , span [className "small"] [text item.description] ] renderApp StateEmpty = p [] [text "Loading..."] renderApp (StateErrored e) = p [className "error"] [text (message e)] renderApp (StateGotItems items) = map renderItem items action Init = do response <- fetch (URL "/api/item") {method: Get} jsonString <- text response itemsOrError <- readJSON jsonString case itemsOrError of Right items -> put (StateGotItems items) Left error -> put (StateErrored error) app = mkComponent { initialState: StateEmpty , render: renderApp , eval: mkEval ( defaultEval { handleAction = action , initialize: Just Init } ) }