npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

@pyreon/reactivity

v0.32.0

Published

Signals-based reactivity system for Pyreon

Readme

@pyreon/reactivity

Standalone fine-grained reactivity primitives — signals, computeds, effects, stores, resources, scopes.

@pyreon/reactivity is the foundation layer every other Pyreon package builds on, but it has zero framework dependencies and works on its own in Node, Bun, edge workers, or any JavaScript environment without DOM or JSX. Subscribers are tracked via an inline _d1 single-subscriber slot that promotes to a Set<() => void> on the second subscriber (the dominant per-<For>-row case pays no Set allocation); batches use a pointer swap for zero-allocation grouping. Two-tier batch flush (computed recompute → effect run) prevents stale reads in diamond-shaped dependency graphs.

Install

bun add @pyreon/reactivity

Quick start

import {
  signal, computed, effect, batch, onCleanup, watch, untrack,
  createStore, createResource, effectScope,
} from '@pyreon/reactivity'

const count = signal(0)
const doubled = computed(() => count() * 2)

const dispose = effect(() => {
  console.log('doubled:', doubled())
  onCleanup(() => console.log('cleaning up'))
})

batch(() => {
  count.set(1)
  count.set(2) // subscribers fire once, with doubled = 4
})

watch(() => count(), (next, prev) => console.log(`${prev} → ${next}`))

const store = createStore({ todos: [{ text: 'Learn Pyreon', done: false }] })
store.todos[0].done = true // fine-grained update, no immer

dispose()

The signal contract

const x = signal(0)
x()           // read (subscribes if inside a tracked scope)
x.set(1)      // write
x.update(n => n + 1)
x.peek()      // read without subscribing

Signals are callable functions, not .value getters (Vue) and not [state, setState] tuples (React). Calling the signal as a function is the read; signal(5) does NOT set the value — it reads and discards the argument. Dev mode warns; the @pyreon/lint rule signal-write-as-call flags it statically.

Optional name for debugging: signal(0, { name: 'count' }) — the @pyreon/vite-plugin injects names automatically in dev.

Computed

const doubled = computed(() => count() * 2)
const sameRef = computed(() => obj(), { equals: (a, b) => a.id === b.id })

Lazy, memoized, auto-tracking. Recomputes only when a dependency changes AND a subscriber actually reads it. Pass a custom equals to dedupe by structural identity instead of Object.is.

Effects

const dispose = effect(() => {
  console.log(count())
  onCleanup(() => console.log('before next run / on dispose'))
})
dispose()

effect() re-runs on tracked-signal change; the returned function disposes. Returning a cleanup function from the effect body is supported; onCleanup(fn) is the explicit form. renderEffect() is a lighter DOM-targeted variant that does NOT support onCleanup and does NOT register with EffectScope — used internally by @pyreon/runtime-dom.

watch(source, callback) is the explicit-source variant: source is evaluated for tracking, callback(next, prev) runs on change, and returning a cleanup function is honored.

Batching

batch(() => {
  count.set(1)
  count.set(2)
}) // subscribers notified ONCE with count=2

batch() defers subscriber notifications until the end of the callback. nextTick(): Promise<void> resolves after the current flush — useful for awaiting DOM updates in tests.

Stores

const store = createStore({ count: 0, todos: [{ text: 'a', done: false }] })
store.count++              // notifies
store.todos[0].done = true // deep — notifies

const shallow = shallowReactive({ user: { name: 'a' } })
shallow.user = { name: 'b' } // notifies
shallow.user.name = 'c'      // does NOT notify (shallow)

const raw = markRaw(thirdPartyClassInstance) // skip proxy

createStore returns a deeply-reactive proxy. shallowReactive proxies only the top level. markRaw opts an object out of proxying — useful for class instances, DOM nodes, third-party objects. reconcile(target, source) patches an existing store to match source without remounting.

Caveat: Map, Set, WeakMap, WeakSet, Date, RegExp, Promise, Error are returned RAW. Mutating them does not notify; assign a new instance to trigger updates.

Resources

const user = createResource(() => userId(), async (id) => {
  const r = await fetch(`/api/users/${id}`)
  return r.json()
})

user.data()      // T | undefined
user.loading()   // boolean
user.error()     // Error | undefined
user.refetch()

createResource(source, fetcher) re-runs the fetcher whenever source changes; stale responses are dropped via an internal request-id guard. Resources created outside an EffectScope must be dispose()-d explicitly to avoid leaks.

EffectScope

const scope = effectScope()
scope.runInScope(() => {
  effect(() => console.log(count()))
  onScopeDispose(() => console.log('scope ended'))
})
scope.stop() // disposes every effect inside

Groups effects for bulk disposal — used internally by @pyreon/runtime-dom's mount pipeline. getCurrentScope() returns the active scope; setCurrentScope(scope) is the escape hatch for advanced cross-tree integrations.

Internal arrays (_effects, _updateHooks) are lazy-allocated — scopes with no effects cost only the object itself.

Selectors

const selected = signal<string | null>(null)
const isSelected = createSelector(() => selected())

<For each={items} by={i => i.id}>
  {(item) => <li class={() => isSelected(item.id) ? 'active' : ''}>{item.name}</li>}
</For>

createSelector(source) returns a function that, when called with a key, only notifies subscribers when the key transitions in or out of the selected state. O(1) instead of N effect runs on selection change.

For the canonical <For> + createSelector className/text-content shape, Selector<T> also exposes a direct .subscribe(key, updater) API that skips the full renderEffect setup — the first-subscriber-per-key path allocates just 1 dispose closure + 1 Map entry (the updater is stored as a bare function, no Set; a Set is only allocated when a 2nd subscriber arrives for the same key):

const dispose = isSelected.subscribe(row.id, (matches) => {
  rowEl.className = matches ? 'selected' : ''
})

The @pyreon/compiler auto-promotes the natural JSX shape class={() => isSelected(row.id) ? 'on' : 'off'} to this fast path — you don't need to call .subscribe directly. See the compiler docs for the bail catalog.

Cell — minimal alternative to signal

import { cell } from '@pyreon/reactivity'
const c = cell(0)
c.get(); c.set(1); c.subscribe(listener)

cell() is a class-based primitive with a single-listener fast path and one allocation per cell. It is not callable and does not participate in effect tracking — use it only for cross-cutting state where the signal-tracking overhead would be wasteful.

wrapSignal — writable side-effect facade

import { signal, wrapSignal } from '@pyreon/reactivity'

const base = signal(load())
// A callable that reads like a signal but persists on every write:
const stored = wrapSignal(base, {
  set: (next) => {
    localStorage.setItem('k', JSON.stringify(next)) // side effect
    base.set(next)                                   // then write through
  },
})
stored()            // read (tracks)
stored.set(value)   // runs your custom set: side effect, then base.set

wrapSignal(base, { set, update? }) is the canonical primitive for "a signal whose writes carry a side effect" (persistence, cross-tab sync, patch emission, validation). It wraps a base signal() with a callable whose reads — (), .peek, .subscribe, .direct — AND the internal _v field delegate to base, while .set runs your custom handler (do the side effect, then call base.set(value) to commit). The _v forwarding is load-bearing: the compiler's _bindText fast path reads source._v directly, so a hand-rolled facade missing it binds '' forever — the exact bug class this primitive retires. update is optional (defaults to set(fn(peek()))). @pyreon/storage (all 5 backends) and @pyreon/state-tree were migrated onto it; the pyreon/storage-signal-v-forwarding lint rule exists precisely because this primitive was missing before.

Debugging

import { setErrorHandler, inspectSignal, onSignalUpdate, why, getReactiveTrace } from '@pyreon/reactivity'

setErrorHandler((err, source) => reportToSentry(err, { tag: source }))

const count = signal(0, { name: 'count' })
onSignalUpdate(count, (next, prev) => console.log('count', prev, '→', next))
inspectSignal(count) // { name, value, subscribers: number }
why(count)           // print dependency graph for this signal

activate/deactivate/getReactiveGraph/getReactiveFires form the opt-in bridge consumed by the Pyreon devtools — zero cost until activated, gated by process.env.NODE_ENV !== 'production', tree-shaken in production.

Documentation

Full docs: docs.pyreon.dev/docs/reactivity (or docs/src/content/docs/reactivity.md in this repo).

License

MIT