@kyneta/reactive

Derived values over Kyneta documents that keep themselves up to date.

reactive(thunk) runs thunk, records exactly which parts of which documents it read, and runs it again when one of them changes. Nothing is compared by value: a computation re-runs when something it read changed, and only then. A burst of changes — a batch, a sync merge — re-runs it once, on the next microtask.

import { reactive } from "@kyneta/reactive"
import { batch, createDoc, Schema } from "@kyneta/schema"

const doc = createDoc(
  Schema.struct({
    todos: Schema.list(Schema.struct({ text: Schema.string(), done: Schema.boolean() })),
  }),
)

const remaining = reactive(() => doc.todos().filter(t => !t.done).length)

remaining.subscribe(() => console.log("remaining:", remaining()))

batch(doc, d => d.todos.push({ text: "Write the README", done: false }))
// next microtask: "remaining: 1"

API

| Export | What it does | |--------|--------------| | reactive(thunk) | A Reactive<T>: call it (r()) or read r.current for the value, recomputed first if something it read changed. r.subscribe(cb) hears each recompute; r.version advances with each. | | computed(thunk) | The same function as reactive, under the name some code reads better with. | | track(feed) | Read a changefeed that is not a schema ref — an @kyneta/index Collection, exchange.peers — as a dependency of the enclosing computation. Schema refs are tracked when you call them. | | diffDeps | The pure subscription diff the runtime uses; exported for bindings. |

A Reactive is itself a changefeed, so one computation can read another, and @kyneta/react's useValue and useTracked render one directly.

A thunk must be pure over what it captures and what it reads through Kyneta. A thunk that reads something untracked (Date.now(), a mutable variable) does not re-run when that changes.

See TECHNICAL.md for how tracking, coalescing and glitch-freedom work.