voithure

@voithure.bsky.social

21 Followers 30 Following

Developer 📍🇫🇷 Paris https://github.com/arthur-fontaine https://x.com/voithure https://arthurfontaine.fr

import { createStateMachine } from "../../../shared/pkg/nanostores-state-machine/nanostores-state-machine"; interface SessionStore { token: string | null; } export const sessionMachine = createStateMachine() .context<SessionStore>({ token: null, }) .events<{ login: { token: string }; logout: undefined; }>() .states({ authenticated: (state) => state .onEntry((context, set) => set({ token: context.token })) .onReceive({ logout: () => "unauthenticated", }), unauthenticated: (state) => state .onEntry((_, set) => set({ token: null })) .onReceive({ login: (_, set, { token }) => { set({ token }); return "authenticated"; }, }), }) .initial("unauthenticated") .stateMachine;
Evan You said Theo - would be much better if you reached out first. Rolldown itself is beta but we haven't really made a proper release for Rolldown-Vite yet so you are testing the wrong target. There are some unfortunate issues with your benchmarks: 1. Rolldown-Vite doesn't need `vite-plugin-react` because JSX handling is built-in. You are basically paying the cost of Babel transform on every module for nothing in your benchmark when it can be done in Rust! 2. Rolldown-Vite ports many Vite-internal JS plugins to Rust, but they are not enabled by default currently. When we do a proper release, we will ask users to opt-in to native mode so we can more gradually test drive and phase it in. When opted-in, Rolldown-Vite is more than 3x faster than current Vite. 3. The bundle size issue is a known issue (de-optimized CJS bindings because React is still... CJS) and has a fix incoming. Relevant PR to your benchmarking repo: https://github.com/t3dotgg/rolldo