Set up SDK

WeaveDB has 4 types of SDKs.

  • a full WeaveDB SDK wrapping the Warp SDK for the web weavedb-sdk
  • a full SDK for NodeJS weavedb-sdk-node

The query APIs for both the SDKs are the same.

Install WeaveDB


yarn add weavedb-sdk


npm install weavedb-sdk

weavedb-sdk-node comes with additional methods on, con, getCache, cgetCache for pub/sub integration.

Instantiate WeaveDB

import WeaveDB from "weavedb-sdk"
const db = new WeaveDB({ contractTxId: WEAVEDB_CONTRACT_TX_ID })
await db.init()

Optional Parameters

common parameters

  • remoteStateSyncEnabled : true | false (default : true)
    To use Warp D.R.E. (opens in a new tab), set it true. D.R.E. works only with weavedb-sdk in browser for now.

  • remoteStateSyncSource : D.R.E. URL (default :
    D.R.E. (opens in a new tab) 1, 2, 3, 4, 5, and 6 are available. Use the https://dre-[NUM] format.

  • network : mainnet | testnet | localhost (default : mainnet)

  • port : port for localhost (default : 1820)

  • arweave : custom arweave-js (opens in a new tab) settings
    e.g. {host: "", port: 443, protocol: "https"}

  • old : true | false (default : false)
    WeaveDB contracts v0.7 and less are not compatible with the latest warp SDK. Set this true to make it work with old DB instances.

  • nocache : true | false (default : true for node, false for web)
    Set the default nocache value. If set false, the SDK returns dryWrite result before sending the tx to Warp. dryWrite is performed on virtual state kept by the WeaveDB SDK, or cached state kept by the Warp SDK without making any http requests, so it's just a matter of milliseconds to return the result.

  • wallet : an admin arweave wallet, note this is different from the default signing wallet

weavedb-sdk-node only parameters

  • subscribe : true | false (default : true)
    Warp subscription plugin (opens in a new tab) needed for on, con, getCache, cgetCache methods, only available with NodeJS

  • onUpdate : function
    A function to execute on pub/sub updates. newState and input object will be passed down.

import WeaveDB from "weavedb-sdk-node" 
new WeaveDB({
  onUpdate: (newState, query, cache, input)=>{}
  • cache : leveldb | lmdb | redis (default : lmdb)
    LMDB has better performance than LevelDB (opens in a new tab) and also is capable of concurrency, but only available with NodeJS. leveldb is what the Warp SDK uses by default, but you can only run one SDK instance with it.

  • lmdb : lmdb settings to merge with the default settings

  state: { dbLocation: "./cache/warp/state" },
  contracts: { dbLocation: "./cache/warp/contracts" },
  src: { dbLocation: "./cache/warp/src" }
  • redis : redis settings
  prefix: "warp", // default to "warp"
  url: "redis://localhost:6379" // default to null

The Redis cache keys will be

  • [prefix].[contractTxId].state.[sortKey]
  • [prefix].[contractTxId].contracts.[sortKey]
  • [prefix].[contractTxId].src.[sortKey]


With onDryWrite option, the SDK returns a virtually calculated result before sending the query to Warp.

You can execute dryRead queries immediately after dryWrite to include in the returned object.

This is great performance optimization to achieve web2-like speed and UX with the smart contract DB.

const result = await db.set({ name: "Bob" }, "ppl", "Bob", {
  onDryWrite: {
    cache: true,
    cb: async ({ nonce, signer, cache, success, duration, error, func, state, results }) => {
	  console.log(`dryRead results: ${results}`)
	  console.log(`Bob: ${results[0].result}`)
    read: [["get", "ppl"], ["get", "ppl", "Bob"]], // an array of dryRead queries
console.log(`regular result: ${result}`)
  • cache: if set true, it will be calculated against the virtual state kept by the WeaveDB SDK, which is much faster (a few ms) than against the Warp SDK dryWrite (a few hundred ms due to checking the latest state with a http request). false is still faster than a regular tx execution process (a few seconds). The difference is true might return the wrong dryRead results if some parallel queries are ongoing on other nodes, which will be solved and roll-backed within 5 seconds.
  • cb: a callback function to immediately execute upon dryWrite.
  • read: read queries to immediately execute against the virtual state after dryWrite. The results come in the write query return object. This is a great performance optimization compared with separate read queries after a write query.

The code below is equivalent to the above, but it will take 3-5 seconds, where as the above takes only around 50 milliseconds.

const result = await db.set({ name: "Bob" }, "ppl", "Bob")
console.log(`regular result: ${result}`)
const Bob = await db.get("ppl", "Bob", true)
console.log(`this will take 3 - 5 sec: ${Bob}`)

with Light Client

onDryWrite can be used with weavedb-client / weavedb-node-client too. But in that case, there is no cb option and the returned value of the entire function will be the result from dryWrite execution. This is because the connection to the gRPC node is a one-off gRPC request, which returns a result only once.

After getting a dryWrite result, you can use getResult() to fetch the finalized result.

const dryWriteResult = await client.set({ name: "Bob" }, "ppl", "Bob", {
  onDryWrite: {
    cache: true,
    read: [["get", "ppl"], ["get", "ppl", "Bob"]], // an array of dryRead queries
console.log(`dryWrite result: ${dryWriteResult}`) // 50-200 ms
console.log(`finalized result: ${await dryWriteResult.getResult()}`) // 3-4 sec