Struct TxBroadcaster 

pub struct TxBroadcaster { /* private fields */ }

Transaction broadcaster responsible for gRPC transmission with retry logic.

Works with TransactionManager to handle sequence mismatch errors gracefully. Uses [RetryManager] with exponential backoff for reliable delivery.

Broadcast Modes

Stateful Orders (long-term/conditional): broadcast_with_retry

Serialized through a semaphore to prevent sequence races. Cosmos SDK requires stateful transactions to have unique, incrementing sequence numbers. The semaphore ensures allocate → build → broadcast happens atomically for each operation.

On sequence mismatch (Cosmos SDK error code 32 or dYdX code 104):

1. The should_retry callback sets a flag indicating resync is needed
2. The RetryManager applies exponential backoff
3. On next attempt, the operation checks the flag and resyncs sequence from chain
4. A new transaction is built with the fresh sequence and broadcast

Short-term Orders: broadcast_short_term

dYdX short-term orders use Good-Til-Block (GTB) for replay protection instead of Cosmos SDK sequences. The chain’s ClobDecorator ante handler skips sequence checking for short-term messages. This means:

• No semaphore needed (fully concurrent)
• Cached sequence used (no increment, no allocation)
• No sequence-based retry logic needed

Implementations

impl TxBroadcaster

pub fn new(grpc_client: DydxGrpcClient) -> Self

Creates a new transaction broadcaster.

pub async fn broadcast_with_retry( &self, tx_manager: &TransactionManager, msgs: Vec<Any>, operation_name: &str, ) -> Result<String, DydxError>

Broadcasts a prepared transaction with automatic retry on sequence mismatch.

Serialization: Acquires a semaphore permit before allocating sequence, building, and broadcasting. This ensures transactions are broadcast in sequence order, preventing “sequence mismatch” errors from concurrent calls.

On sequence mismatch (code=32), resyncs from chain, allocates new sequence, rebuilds the transaction, and retries with exponential backoff.

Arguments

• tx_manager - Transaction manager for sequence resync and rebuilding
• msgs - Original messages to rebuild on retry
• operation - Human-readable operation name for logging

Returns

The transaction hash on success.

Errors

Returns error if all retries are exhausted or a non-retryable error occurs.

pub async fn broadcast_short_term( &self, tx_manager: &TransactionManager, msgs: Vec<Any>, operation_name: &str, ) -> Result<String, DydxError>

Broadcasts a short-term order transaction without sequence management.

Errors

Returns error if building or broadcasting fails.

pub async fn broadcast_once( &self, prepared: &PreparedTransaction, ) -> Result<String, DydxError>

Broadcasts a prepared transaction without retry.

Use this for optimistic batching where you handle failures externally, or when you’ve already prepared a transaction and want direct control.

Returns

The transaction hash on success.

Errors

Returns error if the gRPC broadcast fails.

Trait Implementations

impl Debug for TxBroadcaster

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.