TanStack
TanStack is a set of utilities for building web applications.
TanStack Query is a data-fetching and state management library.
Electric and TanStack
Electric works very well together with TanStack Query, where Electric provides the read-path sync and TanStack provides a local write-path with optimistic state.
In this configuration, Electric and TanStack can provide a fully offline-capable system with active-active replication of both reads and writes.
Example
The example below shows a simple todo application that uses Electric for read-path sync and TanStack for local optimistic writes.
Electric is used to sync a shape. TanStack is used to apply mutations and maintain optimistic state. When a mutation is confirmed, it cleares the optimistic state. When the component renders, it merges the optimistic state into the shape data.
import { getShapeStream, useShape } from "@electric-sql/react"
import {
useMutation,
useMutationState,
useQueryClient,
} from "@tanstack/react-query"
import { matchStream } from "./match-stream"
import { v4 as uuidv4 } from "uuid"
import "./Example.css"
type Item = { id: string }
const baseUrl = import.meta.env.ELECTRIC_URL ?? `http://localhost:3000`
const baseApiUrl = `http://localhost:3001`
const itemsUrl = new URL(`/items`, baseApiUrl)
const itemShape = () => ({
url: new URL(`/v1/shape`, baseUrl).href,
params: {
table: `items`,
},
})
async function createItem(newId: string) {
const itemsStream = getShapeStream<Item>(itemShape())
// Match the insert
const findUpdatePromise = matchStream({
stream: itemsStream,
operations: [`insert`],
matchFn: ({ message }) => message.value.id === newId,
})
// Insert item
const fetchPromise = fetch(itemsUrl, {
method: `POST`,
body: JSON.stringify({ id: newId }),
})
return await Promise.all([findUpdatePromise, fetchPromise])
}
async function clearItems(numItems: number) {
const itemsStream = getShapeStream(itemShape())
// Match the delete
const findUpdatePromise =
numItems > 0
? matchStream({
stream: itemsStream,
operations: [`delete`],
// First delete will match
matchFn: () => true,
})
: Promise.resolve()
// Delete all items
const fetchPromise = fetch(itemsUrl, { method: `DELETE` })
return await Promise.all([findUpdatePromise, fetchPromise])
}
export const Example = () => {
const queryClient = useQueryClient()
const { data: items } = useShape<Item>(itemShape())
const submissions: Item[] = useMutationState({
filters: { status: `pending` },
select: (mutation) => mutation.state.context as Item,
}).filter((item) => item !== undefined)
const { mutateAsync: addItemMut } = useMutation({
scope: { id: `items` },
mutationKey: [`add-item`],
mutationFn: (newId: string) => createItem(newId),
onMutate: (id) => {
const optimisticItem: Item = { id }
return optimisticItem
},
})
const { mutateAsync: clearItemsMut, isPending: isClearing } = useMutation({
scope: { id: `items` },
mutationKey: [`clear-items`],
mutationFn: (numItems: number) => clearItems(numItems),
onMutate: () => {
const addMutations = queryClient
.getMutationCache()
.findAll({ mutationKey: [`add-item`] })!
addMutations?.forEach((mut) => queryClient.getMutationCache().remove(mut))
},
})
// Merge data from shape & optimistic data from fetchers. This removes
// possible duplicates as there's a potential race condition where
// useShape updates from the stream slightly before the action has finished.
const itemsMap = new Map<string, Item>()
if (!isClearing) {
items.concat(submissions).forEach((item) => {
itemsMap.set(item.id, { ...itemsMap.get(item.id), ...item })
})
} else {
submissions.forEach((item) => itemsMap.set(item.id, item))
}
return (
<div>
<div>
<button
type="submit"
className="button"
onClick={() => addItemMut(uuidv4())}
>
Add
</button>
<button
type="submit"
className="button"
onClick={() => clearItemsMut(items.length)}
>
Clear
</button>
</div>
{[...itemsMap.values()].map((item: Item, index: number) => (
<p key={index} className="item">
<code>{item.id}</code>
</p>
))}
</div>
)
}
See the Tanstack example for the full source code.
Help wanted Good first issue
We have an open GitHub issue tracking this if you'd like to contribute a library based on the tanstack
example that integrates Electric and TanStack into a higher level interface.
Please leave a comment or ask on Discord if you'd like any pointers or to discuss how best to approach this.