Wallet's state persistence

[a-ruban-craftsoft]

Hey! I am building some standalone wallet like meta mask or bitcoin wallet that syncs the transactions, blocks, balances with blockchain. I am trying to figure out how it is possible to store current wallet state to file and then to load it on next application start. Is it supported or no? Or ether.js is not that library that can help with it and I should create some own solution?

[ricmoo]

Depending on what you need to restore from or to (e.g. Do you need the mnemonic? Do you need to derive multiple paths from the HD seed? Does your wallet need to preserve the chain ID?) you can either store the mnemonic or just the privateKey. Either is enough to reconstruct the Wallet object.

If storing the HDNode, a read-only extended key might also be useful if you need to generate all child addresses, but not their private keys.

Much of the decisions depend on how your wallet operates and the functionality you wish to provide.

[a-ruban-craftsoft]

Let's take into account only 1 address. Without child derivation.
I want to create wallet using mnemonic. I run the app that creates it, syncs the blocks and transactions. Then the app goes offline for some time and back online and I need to start sync everything from the last synced block. And then I start listening to new blocks, transactions and store all of the updates and last synced block. Bitcoin wallets works like that. But I am not sure can I do something like that with ether.js or no?

[ricmoo]

There is generally only 1 private key associated with an Ethereum wallet (opposed to Bitcoin, which uses multiple account indices with internal and external accounts for change and receiving respectively).

If you use Wallet.createRandom() the wallet will have a mnemonic, which you can present the user to backup. It’s up to you at this point whether you discard the mnemonic for security and store only the private key or retain it so additional accounts can be added in the future. Keep in mind all private data should be encrypted.

For syncing, you will need to retain the last synced blockNumber and you will probably need to use an API similar to Etherscan to find all transactions. Keep in mind that there is no equivalent function in Ethereum to find all transactions, since it is not UTXO-based, the only data required for a transaction is the nonce, so one number per account is stored in the Ethereum blockchain that is indexed.

Think of how a Bitcoin wallet requires a backend with the entire UTXO set, indexed by the hash160 addresss (I haven’t done BTC dev in a long time, between SegWit and bech32, there may be more indexing required). You need an equivalent backend to pull the transactions for a given address. This is not built-in, but with an archive node may be available using the trace and debug API’s, but it is likely not performant; so you will want to index these.

This is the exact API that Etherscan provides, but it is highly throttled and you would likely want to cache this. If you are building a commercial product, you will want to reach out to them to see if you can purchase an api key with higher request rates or build out your own indexed dataset.

Also, accounting for orphans, you will probably want to pull from several blocks earlier than the last pulled block.

Hope that helps. :)

[a-ruban-craftsoft]

Thanks a lot!