sign_typed_data_v4 TODO: is the eth_signTypedData_v4 override still needed?

[teyrebaz33]

Describe the question

crates/seismic/rpc/src/eth/ext.rs's sign_typed_data_v4 handler carries an author TODO:

/// Handler for: `eth_signTypedData_v4`
///
/// TODO: determine if this should be removed, seems the same as eth functionality
async fn sign_typed_data_v4(&self, from: Address, data: TypedData) -> RpcResult<String> {
    debug!(target: "reth-seismic-rpc::eth", "Serving seismic eth_signTypedData_v4 extension");
    let signature = EthTransactions::sign_typed_data(&self.eth_api, &data, from)
        .map_err(|err| err.into())?;
    let signature = alloy_primitives::hex::encode(signature);
    Ok(format!("0x{signature}"))
}

What I found

Upstream reth's own core.rs (reth_rpc_eth_api) implements eth_signTypedData (no _v4 suffix) and returns raw Bytes:

/// Handler for: `eth_signTypedData`
async fn sign_typed_data(&self, address: Address, data: TypedData) -> RpcResult<Bytes> {
    trace!(target: "rpc::eth", ?address, ?data, "Serving eth_signTypedData");
    Ok(EthTransactions::sign_typed_data(self, &data, address)?)
}

So the two aren't actually the same "eth functionality":

• Method name: standard reth exposes eth_signTypedData; this override additionally exposes eth_signTypedData_v4. greping the crate, eth_signTypedData (no _v4) is never separately registered here — only the _v4 name is served.
• Response shape: standard returns raw Bytes; this override returns a hex String prefixed with 0x.

_v4 is the name real wallets and dapp libraries actually call — MetaMask's docs recommend it for EIP-712 signing, and other client libraries (e.g. seaport-js) explicitly fall back between eth_signTypedData_v4 and eth_signTypedData because node/wallet support for the two names differs. Removing this override would mean the node no longer answers eth_signTypedData_v4 at all, breaking anything that calls it by that name specifically.

Suggested resolution

Doesn't look like it should be removed. I think the TODO can be resolved by replacing it with a comment explaining the naming/response-shape difference (so a future reader doesn't have to re-derive this), rather than removing the handler. Happy to open a doc-only PR for that if useful.